Operational aspects of signal-Averaged electrocardiography

Operational aspects of signal-Averaged electrocardiography

Operational Aspects of Signal-Averaged David N. Dunbar IGNAL-AVERAGED electrocardiography (SAECG) is a computerized processing technique that greatl...

2MB Sizes 0 Downloads 77 Views

Operational

Aspects of Signal-Averaged David N. Dunbar

IGNAL-AVERAGED electrocardiography (SAECG) is a computerized processing technique that greatly enhances the resolution of the surface electrocardiogram so that lowamplitude cardiac potentials can be recorded from the body surface. Initially, this technique was used to record His-Purkinje potentials from the body surface to evaluate atrioventricular conduction.‘,’ It is now principally used to record late potentials to identify patients at increased risk for life-threatening ventricular arrhythmias. Late potentials are low-amplitude highfrequency signals located at the end of the QRS and the beginning of the ST segment. Late potentials were first identified by direct epicardial recordings in animal experiments.3-5 These studies showed that local electrical activity in the ischemic zones persisted well beyond the end of the QRS complex. The continuous electrical activity bridged the diastolic interval between the initiating and the re-entrant ventricular ectopic beats. Berbari et al first used SAECG to record these low-amplitude cardiac potentials from the body surface, which extended into the ST segment and corresponded to delayed epicardial potentials in the ischemic zone of an animal model.” Josephson et al have shown continuous local electrical activity from endocardial recordings during ventricular tachycardia (VT) in man and have noted that markedly prolonged and fractionated electrograms were also present in these patients during sinus rhythm.’ Fontaine et al,” Breidthardt et al,” and Uther et al’” reported visual identification of late potentials in patients who had VT. Simson et al used SAECG with bidirectional digital high-pass filters to detect and quantify late potentials, thus identifying patients who had VT after myocardial infarction (MI).” Many investigators have found that late potentials are present in 60% to 90% of patients who have a history of sustained VT or ventricular fibrillation (VF).“m’” The fractionated electrograms responsible for late potentials have been shown in animal studies to rise from widely separated

S

progress

in Cardiovascular

Djseases,

Vol XXXV,

No 5 (March/April),

Electrocardiography

and Pablo Denes

islands of muscle fibers embedded in the fibrous tissue of healing MI with delayed asynchronous conduction due to marked fiber disorientation.” This review will discuss the operational factors that are important in the recording, processing, and interpretation of SAECG in the time domain. The SAECG parameters used to identify the late potentials and the range of normal values will also be discussed along with the indications and limitations of SAECG recording, specifically addressing its application for nonischemic cardiomyopathy. METHODS

Technical Factors Influencing Recording of Late Potentials Late potentials are rarely recorded by standard electrocardiography because their low amplitude (usually < 10 FV) causes the late potentials to be obscured by background noise from a variety of sources. Electromagnetic interference is a common source of noise in the hospital environment from multiple electrical power sources. Shielded rooms or Faraday cages have been used for high resolution electrocardiography to attenuate electromagnetic interference but are impractical because of their immobility.“? Electromagnetic interference can be decreased by the design of the SAECG equipment using shielded cables, instrument grounding, high-resolution band-pass filters, and highresolution pre-amplifiers. Notch filters can be used to filter out 60 Hz alternating current interference along with its harmonics, but this is not recommended.23 Background noise can also occur at the electrode skin interface that is related to electrode instability, polarization, or high impedance. Skeletal muscle potentials are From the Cardioloa Division, Hennepin County Medictrl Center, Minneapolis, and the Section of Cardio&y, St Paul Ramsey Medical Center, St Paul, MN. Address reprint requests to David N. Dunbar, MD, Curdiologv Division. Hennepin County Medical Center, 701 Park Ave. Minneapolis, MN 55415. Cop.vright 8 1993 by W.B. Saunders Company 0033-0620/9313505-0002$5.00/O

1993:

pp 329.348

329

330

a principal source of background noise. Filtering techniques are largely ineffective in attenuating skeletal muscle potentials because their high-frequency spectra overlap those of cardiac potentials, especially the high-frequency late potentiaIs.‘4 The random nature of the background noise relative to the underlying cardiac signals allows SAECG techniques to greatly enhance the cardiac signal-to-noise ratio. Multiple samples of the repetitive QRS wave form are amplified and recorded with band-pass filtering from 0.05 to 250 Hz. The analog signals are converted to a digital format with a sampling rate of 1,000 to 2,000 Hz. The QRS signals are then compared using a template recognition algorithm to identify sinus beats and reject noisy or ectopic beats. The sinus beats must be precisely aligned by identifying a unique feature within the QRS complex, the fiducial point, before the signal amplitudes are summed and averaged. Variability in QRS alignment (jitter) causes the signal to be reduced without corresponding reduction in random noise. Increasing jitter causes greater signal attenuation at higher frequencies.” The signal averaging process reduces random noise proportional with the square root of the number of cycles recorded, such that recording 100 to 400 sinus beats reduces the background noise by lo- to 20-fold. Commercially available devices have a resolution of 1 FV. This is well beyond the resolution of standard electrocardiography (1 uV = 1/lOO mm at standard gain of 1 mV = 10mm). Optimal noise reduction begins with patient preparation to limit the noise that is present during signal acquisition. Skeletal muscle noise can be decreased by recording while the patient is warm, quiet, and relaxed in the supine position. Noise from the electrode skin interface can be reduced with proper electrode preparation. The electrode site is lightly abraded with fine sandpaper and wiped with alcohol to rcduce skin impedance. Silver-chloride electrodes are recommended because of lower impedance, lower offset potentials, and less polarization than other metal electrodes.‘” Contact between the electrode and skin interface through a layer of electrolyte gel enhances lead stability and minimizes noise.

DUNBAR

AND

DENES

Parameters Used for Late Potential Detection and Dejinitions

SAECG to detect late potentials has used various devices using different surface electrode positions, filtering techniques, and parameters to define late potentials. Breithardt et al developed a method of visually identifying late potentials from bipolar recordings taken from four precordial electrodes.’ The amplified signal was processed with a high-pass filter at 100 Hz. Late potentials were initially identified by signals that extended more than 10 milliseconds beyond QRS offset. Criteria were later modified to identify low-amplitude potentials lasting more than 10 milliseconds in duration irrespective of the surface QRS duration.” Uther et al developed the signal averaging technique for the determination of ventricular activation time from orthogonal X, Y, and Z leads using a high-pass filter of 0.05 Hz.“’ The ventricular activation time is determined by the earliest QRS onset and the latest QRS offset in any lead. The onset and offset were defined by a value of twice the baseline noise. Patients who have sustained ventricular tachyarrhythmias have significantly longer ventricular activation time than normalsiJ Simson introduced a quantitative approach based on computer algorithms to identify QRS end points, amplitudes, durations. and background noise levels from bipolar orthogonal X, Y, and Z lead recordings.” The amplified signals are averaged and processed through a bidirectional high-pass filter. The averaged filtered X, Y, and Z signals are then combined into a vector magnitude that equals the square root of X2 + Y’ + Z’. The resulting vector magnitude (also called filtered QRS) is analyzed with computer-assisted measurements to identify late potentials. The vector magnitude duration has been termed filtered QRS (fQRS, QRS, or QRSd). Late potentials increase the duration of the vector magnitude by delaying QRS offset. Late potentials decreased the amplitude of the last 40 milliseconds of the filtered QRS complex measured in root-mean-square voltage (RMS or RMS-40). Denes et al later reported that the duration of the low-amplitude signal at the end of the filtered QRS could also be used as a discriminator to identify late

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

potentials.‘” This was defined as the duration of the signal under 40 mV in amplitude at the end of the filtered QRS termed low-amplitude signal (LAS). An abnormal SAECG with a late potential is depicted in Fig 1. Lead Systems for Signal Averaging

The results of SAECG are lead-dependent with variability in late potential criteria that is applicable from one lead system to another. Atwood et al found that a precordial bipolar electrode configuration yielded significantly lower RMS voltage than bipolar orthogonal leads whereas filtered QRS duration tended to be equivalent. 26 Theoretically, late potentials should be reasonably well detected with orthogonal leads because late potentials have a dipolar distribution with single positive and negative regions that are determined by body surface mapping.” However, Berbari et al found that signal averaging from an orthogonal lead system may underestimate the duration and occasionally miss late potentials which are identifiable using signal averaging of a 24-lead precordial map.‘” The use of orthogonal bipolar X, Y, and Z leads allow signal recordings with a minimal number of leads. The American Heart Association statement on standards for analysis of ventricular late potentials using SAECG’” has recommended the following position: The X lead should be positioned at the fourth intercostal space in both midaxillary lines. The Y lead should be positioned on the superior aspect of the manubrium and on either the upper left leg or left iliac crest. The Z lead should be placed at the fourth intercostal space (V2 position) with the second electrode positioned directly posteriorly on the left side of the spine. The positive electrodes are at the left, inferior, and anterior. High-Pass Filter Characteristics Frequency)

(Corner

Most investigators have used digital high-pass filtering to aid in the separation of highfrequency late potentials from background noise. Cardiac potentials in the ST and T wave generated by repolarization may reach 100 FV in amplitude and could obscure late potentials. These low-freouencv notentials are areatlv atten-

331

O-250Ht OOmm/81 .OOmm/yV

I

1 I

ector ' iagnityle

Durations [msl: 'Std QRS 76 Total OAS 113 'Under 4OuV 56

kMS Voltages ITotal Last ’ NOiBe

I

WI

QRS 74.7 40ms 5.0 0-i

I

-100

u\

I

I I

I

LIAS 1

-50

-40

uv

uv

Fig 1. Abnormal SAECG from a 68-year-old man who had remote inferior Ml, nonsustained VT, depressed left ventricular function, and inducible sustained monomorphic VT. Note the presence of a late potential with RMS-40 amplitude of 5.0 pV (stippled area) and prolonged LAS duration of 56 milliseconds. The filtered QRS duration of 113 milliseconds is near the upper limits of normal but is much longer than unfiltered (standard GRS) duration of 76 milliseconds. (Recorded with Arrhythmia Research Technologies 1200 EPX (Austin, TX] at corner frequency 40 Hz.)

uated by high-pass filtering. Thus, the signal-tonoise ratio is enhanced for the higher frequency late potentials by high-pass filters in the freauencv ranges of 20 to 100 Hz.

332

Digital high-pass filters are subject to ringing of the filter which produces artifact after abrupt transient high-amplitude signals. Ringing of high-pass filters after the high-amplitude QRS signals could obscure the low-amplitude late potentials occurring early within the ST segment. Simson was first to advocate bidirectional processing of the QRS to prevent filter ringing at QRS offset.” With this process, QRS onset is identified by analyzing the initial portion of the QRS in forward direction to the mid-QRS. The terminal portion of the QRS complex is then analyzed backward from the ST segment to the mid-QRS to identify the QRS offset. This type of high-pass filter provides good attenuation of low-frequency potentials but it does introduce discontinuity in the middle of the QRS complex. Other types of digital filters, such as spectral window or finite impulse response, preserve QRS morphology but may produce phase delay resulting in spreading of the QRS in time.?” The optimal high-pass filter to identify late potentials has not been determined. Most of the spectral power of the electrocardiogram lies at a level less than 40 Hz with small amounts of power spectrum up to 500 Hz.‘~ The spectral domain of late potentials defined using frequency domain (Fast Fourier) analysis shows increased spectral power from 20 to 50 Hz frequencies in patients who have a history of ventricular tachyarrhythmias.3”.3’ Several investigators have examined the effect of variable high-pass filters on quantitative signal-averaged parameters used to identify late The effects of high-pass filpotentials. 13.1x~32~33 ters on QRSd, RMS-40, LAS, and noise were evaluated using multiple high-pass filters ranging from 10 to 100 Hz in normal subjects and patient populations who had heart disease.18.” Both of these studies have shown a progressive and nonlinear decline in total RMS/QRS voltage and the RMS voltage of the last 40 milliseconds as high-pass filtering was increased from 10 to 100 Hz. These studies have also shown a progressive linear increase in LAS duration in both normal subjects and patients who had heart disease. Residual noise levels have shown a marked decline with increasing high-pass filters from 10 to 20 Hz and a more gradual decline in noise from 20 to 50 Hz. The QRSd

DUNBAR

AND

DENES

increased as high-pass filtering was increased from 10 to 20 Hz because the algorithm defining QRS offset is a function of residual noise. Thus, with a much higher noise level at IO Hz, the QRS offset was identified at a significantly higher amplitude (three times the noise level), thus shortening QRS duration. The high noise level at 10 Hz high-pass filtering could also obscure low-amplitude late potentials. With filters from 50 to 100 Hz, there was little change in residual noise. QRS duration tended to plateau at high-pass filters ranging from 20 to 40 Hz and decreased with high-pass filters with frequencies >40 Hz reflecting a greater decrease in RMS voltage than residual noise level. Thus, high-pass filter settings < 20 Hz were not found to be useful for late potential identification because of the excessive noise.‘“J’ With high-pass filters of 100 Hz, the RMS-QRS voltage and RMS-40 were greatly attenuated and LAS values increased. The loss of RMSQRS voltage at high-pass filter settings > 60 Hz prevented LAS analysis in 1% to 13% of patients as filtering settings were increased from 60 to 100 Hz because no portion of the QRS signal exceeded 40 l.r,V in amplitude.” Gomes et al found that the RMS-40 of the 80-Hz filter setting had the highest sensitivity (88%) but a lower specificity (69%) in differentiating patients who had organic heart disease from those who did not have spontaneous VT.lX At 25 Hz, RMS-40 had a high specificity (90%) but a low sensitivity (42%). At 40 Hz, RMS-40 had intermediate values with a sensitivity of 61% and specificity of 88%. In this study, LAS and QRS values had lower sensitivity and specificity than the RMS values of the last 40 milliseconds. The requirement of two variables to define late potentials could enhance specificity (range 86%) to 90%) at the expense of sensitivity (range 18% to 48%). Denes et al also found increased sensitivity in identifying patients who had sustained VT as high-pass filters increased from 25 to 40 Hz.” Caref et al found that LAS at 25 to 40 Hz and RMS-40 at 40 Hz had the best sensitivity in identifying patients who had inducible sustained VT from a patient population with nonsustained VT.” RMS-40 values at frequencies < 20 Hz or > 60 Hz were not good discriminators. Vatterot et al found

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

that QRSd at 25Hz filtering and RMS-40 at 40-Hz and 80-Hz filtering were the SAECG variables that best distinguished those patients who had inducible VT at electrophysiological testing from those who did not have inducible VT.“” The addition of a 250-Hz low-pass filter does not substantially change late potential pararneters.“4 Thus, the SAECG parameters for late potential identification and the values for normal subjects are dependent on high-pass filter frequency settings. These studies suggest that analysis of the SAECG using multiple high-pass filters may improve the predictive accuracy of the SAECG. Residual Noise and Reproducibility

Residual noise is the level of baseline noise that is present after noise reduction techniques through signal averaging. Residual noise is measured in diastolic segments of variable length either before or well after the QRS interval. The residual noise amplitude is dependent on the noise present during signal acquisition, the number of beats averaged, and the filter settings. The signal averaging process reduces noise level proportional to the square root of the number of beats averaged. Typically, 200 to 400 beats are averaged for a noise reduction of 14- to 20-fold over baseline. Residual noise also decreases as the corner frequency increases. The residual noise level can have important effects on determining QRSd and detection of late potentials. The filtered QRS onset and offset is defined in automated systems as the point where voltage exceeds 2% to 3 times the residual noise level. Steinberg and Bigger found that reducing residual noise levels from 1.0 PV to 0.3 yV improved the sensitivity of late potential detection from 46% to 69% in a patient population with sustained ventricular tachyarrhythmias.j5 They recommended that the signalaveraging process should be performed to a prespecified low residual noise level rather than being based on a set number of beats. The American Heart Association task force on SAECG has recommended that residual noise be measured over a 40 millisecond period in the ST or T segments for a 4-pole bidirectional Butterworth filter.‘” They have recommended that the residual noise be < 1.0 FV using a

333

25-Hz high-pass filter and <0.7 FV with a 40-Hz high-pass filter. Signal averaging to a specified low residua1 noise level produces excellent reproducibility.“6J7 Commercially Available Devices

The four most commonly used commercially available SAECG devices in the United States are reviewed in Table 1. Each device uses cross correlation matching QRS complexes with a template, a signal averaging algorithm to reduce noise, and sums X2, Y2, and Z? bipolar lead voltage to generate a vector magnitude by Simson’s method.” These devices differ in the algorithms and the filters that are used to identify residual noise and QRS end points. Each device has or soon will have capabilities for data analysis in frequency domain and spectral temporal mapping in addition to time domain analysis. Little data is available to allow comparison of commercially available devices. Oeff et al compared SAECG results using four devices: three devices involved visual identification of late potentials and one device used computer algorithms developed by Simson.j8 Late potentials were detectable in 12% to 21% of patients. All four devices gave corresponding positive results in 5.5% and corresponding negative results in 68.8% of patients. The SAECG devices yielded differing results in 25.7% of patients, primarily because of differences in methodology for determining filtered QRS offset. Henkin et al compared RMS, fQRS, and LAS parameters recorded from four commercially available SAECG devices and found that up to 19% of the recordings were discordant in at least one SAECG parameter.3Y The RMS-40 value was by far the most variable, showing a marked sensitivity to slight changes in QRS offset. Averaged unprocessed data files from 80 subjects were processed by computer algorithms with three devices to isolate variables due to analysis from those of data acquisition. RMS-40 results were discordant in 23%, QRS in ll%, and LAS in 15% of analyses. Other studies have shown significant differences in SAECG parameters recorded on different SAECG devices.34,4”

DUNBAR

334

Table Parameter

rate

1,000

Filter

Template

Sum of differences 20 points over ms noise

analy-

of 100

segment* fQRS

Midpoint

offset

12-16

bit

2,000

Hz

1,000

Cross

correlation

points

over

80 40 ms

Average unfiltered (0.05300 Hz) voltage in fixed 50.ms

of 5-ms winamplitude

Midpoint tive”

of “adap2 ms window

with amplitude threshold ratiot

of earliest

unfil-

tered QRS In X, Y, or Z lead

Analysis

Time FFT

temporal

Reanalysis option: user defined QRS

Yes

Advantages

Portable Extensive

map

perience

Average RMS voltage of the 2 most quiescent 50 ms windows

Average RMS of the most quiescent 30.ms window in ST

correlation

21

shifts over 128-ms spectral window

segment

of 5 ms win-

Endpoint

of 10.ms

window with average amplitude 2 x RN

in

as QRS offset

Slmtlar

domain

FFT Spectral

to QRS offset

map

domain

Spectral

Time FFT

temporal

application No distortion

to ORS offset

Small,

map

domain

Spectral temporal in development

map

Yes

Yes

flexibility research

Similar

except amplitude 3 A RN

Time FFT

temporal

Precision, especially

ex

Cross

to a fixed

Yes

clinical

Cross correlation 60 points over 60 ms

dow with amplitude 2.7 x RN

=

Hz

Spectral

but uses ratio to RN in 20 ms PR segment window Time

domain

Spectral

Same

1,000

Hz

Btdirectional Butterworth

Midpoint

30.ms RN period ST segment Onset

16 bit

In ST segment

relative

fORS onset

Marquette

16 bit

window

dow with ofRN+3SD

Devices Fidelity

Spectral BispecS

Average RMS voltage in most quiescent 40 ms window in ST

sis

SAECG

DENES

-

Hz

Bidirectional Butterworth

Residual

Available COKWOlllX

16 bit

A-D Conversion Sampling

1. Commercially

Arrhythmia Research Technologies

AND

portable

Less expensive

Versatile incorporated into 12 lead ECG machine with vectorcardiography

of cen-

tral ORS

cost

1200

EPX $23,250

1200

LPX $11,950

Predictor Predictor

$25,500 II $18,500

LP 300-$11,900 LP-Plus-$17,800

MAC-15 ment lution

ECG equipwith high reso option-

$15,000 Abbreviations:

RN, residual

*Model lOl-Average tproprietary information. SCorazonix,

noise

noise; reported,

RMS,

root 1200

mean EPX only

square;

ECG, electrocardiogram.

standard

deviation

of noise

reported.

Inc.

Thus, SAECG parameters, particularly terminal RMS, are exquisitely sensitive to the determination of QRS offset. Commercial devices that vary in the methods that are used to determine QRS offset and residual noise can yield discordant results. Figure 2 depicts an example of SAECG variability due to changes in QRS offset.

Normal

SAECG

Values

The values of SAECG parameters obtained in normal subjects by different investigators are listed in Table 2.13.18,223,41-43 The values for the filtered QRS duration, low-amplitude signal and energy content of the last 40 milliseconds of the QRS at 25, 40, and 80 Hz high-pass filter settings are listed. In the studies for which the

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

335

potentials by the signal-averaging process (Dunbar, Denes: unpublished observations). The mechanism of the underlying ventricular arrhythmia (ie, automaticity or reentry) may influence the prevalence of late potentials. Effects of Intraventricular Disturbances

Fig 2. Two SAECGs recorded 15 minutes apart in a patient using same electrocardiogram lead position. A slight change in the ORS offset determined by computer algorithm results in large changes in RMS40 amplitude (fQRS 91-98 milliseconds and RMS-4040.7-15.6 pV, respectively). (Recorded with Arrhythmia Research Technologies 1200 EPX at corner frequency 40 Hz.)

data was available, the prevalence findings are shown. CARDIAC DISEASE-RELATED INFLUENCING RECORDING POTENTIALS

of positive

FACTORS OF LATE

The ability of the SAECG to identify patients who have sustained VT or ventricular fibrillation (VF) may be limited by factors related to the underlying cardiac diseases and anatomical site of reentry. For a late potential to be detected, it must be of sufficient amplitude and it must extend beyond the QRS complex into the ST segment. Simultaneous depolarization of normal myocardium would obscure the highfrequency, low-amplitude potentials occurring within the QRS complex. Thus, late potentials may not be identifiable if they are of short duration, if they arise from the anterior septum which is normally activated earlier than the posterior basal left ventricle, or if they arise in the presence of bundle branch block.44 The amplitude of the late potential must be large enough to discriminate from n0ise.j” Atria1 fibrillation may produce factitious late potentials through inadequate attenuation of atria1

Conduction

Conduction disturbances, particularly left bundle branch block (LBBB) and QRS duration, have long been recognized as prognostic indicators of an increased risk of sudden cardiac death particularly after MI.4s-Jy However, most SAECG studies have excluded patients who have bundle branch block or marked intraventricular conduction delay because of the observations that delayed ventricular activation could obscure late potentials.44,5” Buckingham et al developed separate SAECG parameters for patients who have normal conduction, patients who have fascicular block or intraventricular conduction delay, and patients who have bundle branch block.S1 Longer fQRS and LAS durations and lower RMS-40 amplitudes were used to define abnormal SAECG parameters in patients who had conduction disease. The criteria provided only limited sensitivity of 62% and specificity of 63% in detecting patients who had sustained VT from a heterogeneous population with organic heart disease and conduction system delays, In contrast, Lindsay et al found that with frequency domain analysis the SAECG was able to identify patients who had sustained VT with a sensitivity of 95% and specificity of 80% despite the presence of bundle branch block.“’ These results, using frequency analysis, were similar to the sensitivity and specificity in identifying patients who had VT without underlying conduction system disease. Haberl et al also reported that frequency domain analysis of the SAECG could identify patients who had bundle branch block and sustained VT after MI.” Criteria are not currently well established for identification of patients who have sustained VT in the presence of bundle branch block using time domain analysis. Effect of Infarct Location The incidence of late potentials is significantly higher in patients who have inferior MIS

336

DUNBAR

Table Author

25 Hz High

Age Iv4

42

34 k 2

GomesT8

25

Flower+

67 33

100

34+

10

FP PA)

i 1.4 c 120

Subjects LAS (msl

94.4

2 10.1
95.9

27 91

11

Freedmanq2

19

30 2 4

Danford

32

25 k 4

23.3

i

42

GomesJ8

33

Danford High-pass

Caref

s3

67.1

1.6

117

20 t 0.7

87.4

k 7.7

96.9

i- 60.3 225

53.7

2 25.2

18.8

20.4

i 103 -25

-

+ 6.8 ~32

-97.3 + 9.4 M <117F
2 5.2 >25

20 + 6 ~32

k 7.2

87.5 88.9

<35

34 f 2

25

34 t 10

100

31 t 9

93.7

i 1.4 cl20

29.5

922 11 \I14

+ 54.9 .’ 20

92.4

27

28 2 4

25

34 2 10

94.2

+ 1.1 .: 39

10

41.6

-+ 9.5

24.1

81 i- 65 >20

2 7.3

58.4

i 38.7 216

58.0

i 32.6

539 zt 12.6

24.9

c 3.5 -1 20

26 + 6 <38

ill1

Games’*

21

Filter

Denes”

80.Hz

+- 1 1 230

942 IO Cl14

31 39

36+

Pass

95.9

93.8

55

Caref

in Normal

QRS (ms)

28 + 9

POW’

40 Hz High

Parameters

DENES

Pass Filter

Denes13

Caref

N

2. SAECG

AND

k 11 .o

filter

100

31 2 9

Abbreviations: FP, false positives; QRS. amplitude of last 40 ms of vector magnitude;

87 2 10

30 + 6

47 2 54

c 107

x42

-17

87.9

t 9.0 5106

filtered QRS duration; M, male; F, female.

(58% to 62%) than in patients who have anterior wall MIS (31%-34%).“‘,54 Because the basal portions of the myocardium are activated late within the QRS complex, low-amplitude signals are more likely to extend beyond the end of the QRS complex in the patients who have inferior MI than in the patient who has anterior MI. Gomes et al found the SAECG has a high sensitivity (100%) but a low specificity (50%) in predicting arrhythmic events after inferior MI and had a high specificity (80%) but lower sensitivity (75%) in patients who have anterior MI.54 Buxton et al found that in patients who have inferior MI and nonsustained VT, the SAECG was a sensitive indicator for inducibility of sustained VT, but in similar patients who had anterior MI it was not useful in predicting inducibility.5-s

32.2

i 8.7

29.0

<50 LAS,

low-amplitude

signal

I 15.7 ?9

duration;

RMS-40,

root-mean-square

Effect of E\volution of Late Potentials Durirlg Acute MI Abnormalities in the SAECG can develop very early in the course of acute MI, but the significance of these abnormalities in predicting early VT or VF is uncertain. McGuier et al reported a 32% incidence of late potentials at presentation with a mean time to SAECG recording of 12.4 -+ 6.6 hours after the onset of chest pain..5h Eight of 21 patients who had late potentials developed early VT and VF compared with 2 of 35 patients who did not have late potentials. In this study, the presence of late potentials on the initial SAECG identified patients who had early VT and VF with a sensitivity of 80% specificity of 72%, and a negative predictive value of 94%. In contrast, El Sherif et

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

al found that only one patient of eight who developed VT or VF in the first 48 hours had late potentialss7 Grimm et al reported a 19% incidence of late potentials at presentation with acute MI but the initial recording did not predict in-hospital VT or VF or mortality.58 A second SAECG during infarct evolution showed that 20% of patients changed categories (abnormal to normal SAECG or vice versa). Hong et a159 found that patients who had abnormal SAECGs performed in the first 2 to 3 days after MI had a significant increase in mortality and arrhythmic events in the first 10 days after infarction. The study by El Sherif et al showed the dynamic nature of the SAECG early after MI and also investigated its value in predicting arrhythmic events during the subsequent year.57 SAECGs were recorded in 156 patients during phase I-O to 5 days, phase II-6 to 30 days, and phase III-31 to 60 days after acute MI. Fiftyone patients (31%) had an abnormal SAECG during one of these three phases. The SAECG was abnormal in 17% of patients in phase I, 25% during phase II, and 18% during phase III. Sixty-nine percent of patients changed SAECG categories from normal to abnormal or vice versa. Twenty-six of 51 patients (50%) developed delayed appearance of late potentials after the first 5 days postinfarction whereas 13 of 51 patients (25%) had a disappearance of late potentials 30 days after MI. The abnormal SAECG during phase II had the most significant relationship to arrhythmic events during the first year (P = 0.0001). Nine of 12 patients who had VT or VF during the first year of follow-up had an abnormal SAECG during phase II. Of note, 7 of the 12 arrhythmic events occurred between 6 and 30 days after infarction. The investigators believed that the optimal time for SAECG recording to determine prognosis was 6 to 30 days, preferably 6 to 14 days after MI. Elgect of Thrombolysis Eldar et al examined the effect of thrombolysis on the evolution of late potentials in the first 10 days after MI. 6oThe incidence of late potentials was similar in the first 2 days after infarction in the group treated with thrombolytics (14% incidence of late potential), and the group

337

treated conservatively (11.8% incidence). There was no change in the incidence of late potentials in the group treated with thrombolysis, however, the incidence of late potentials increased significantly in the control group from 11.8% to 22.5%. Gang et al also found a decreased incidence of ventricular late potentials after successful thrombolytic therapy for acute MLhl In this study, late potentials appeared in two of 44 (5%) patients who were treated with thrombolysis compared with 14 of 62 (23%) in those who were treated conservatively. In the patients who were treated with tissue plasminogen activator (tPA), none of 38 patients who had a patent infarct-related artery had late potentials, but two of six patients who had an occluded infarct artery had late potentials. Multivariate analysis showed that patency of the infarctrelated artery and treatment with tPA were the only independent predictors of the presence of late potentials; whereas, MI location, peak creatine kinase level, coronary artery disease score, and left ventricular ejection fraction were not. Preliminary data from the GISSI-2 study reported a 16.5% incidence of late potentials in patients receiving thrombolytic therapy compared with a 28.9% incidence in patients who did not receive thrombolytic therapy.6’ Lange et al found that patients who had antegrade flow in the infarct-related artery had a significantly lower incidence of late potentials, 4/49 (8%). than those patients in whom the infarct artery remained occluded, 24/60 (40%).6’ Thus, treatment with thrombolytic agents seems to decrease the incidence of ventricular late potentials and may have a long-term benefit in the reduction of spontaneous arrhythmic events. Effect of Arrhythmia Presentation: VF Versus VT Patients presenting with sustained VT have a higher incidence of late potentials (70%-80%) than patients presenting with VF (7%57%).42,64-69Patients presenting with VT have significantly lower RMS amplitudes42,6h and longer LAS69 than patients presenting with VF. The lower incidence of late potentials in patients who have VF parallels the lower incidence of inducible monomorphic VT with programmed electrical stimulation compared with patients presenting with spontaneous VF.42,65.66 The presence of late potentials in patients who

DUNBAR

338

have VF, however, does not predict VT inducibility.42J’6 These results suggest that the substrate for patients who have spontaneous VF is different from that in patients presenting with VT. This identifies a potential limitation of SAECG in detecting patients at risk for faster rates of VT or VF compared with slower and presumably more hemodynamically stable forms of VT. Effects of Antiarrhythmic Drugs Amiodarone, quinidine, and procainamide produce slight prolongation of filtered QRS ranging from approximately 5 to 15 milliseconds or approximately 10% prolongation over baseline.‘O-‘* Mexiletine and metoprolol have no significant effects on filtered QRS duration.7’,73 Class IA and IB antiarrhythmic drugs and amiodarone did not significantly change RMS-40 amplitude or the incidence of late potentials.71.72Simson reported slight prolongation of LAS duration 3.5 & 13 milliseconds, a 6.5% change with antiarrhythmic therapy.‘” Kuchar et al reported that LAS duration correlated with VT cycle length and that prolongation of LAS parallels increases in cycle length with drug therapy. 74 Borbola and Denes evaluated the effect of amiodarone loading therapy on SAECG recordings in patients who had VT.77 They showed that amiodarone prolonged LAS duration 19 milliseconds (61% change) relatively more than fQRS 20 milliseconds (18% change). These findings suggest that amiodarone has a more pronounced effect on the reentrant pathway than on the remaining myocardium. The SAECG does not predict antiarrhythmic drug success in suppressing VT.70.7”,74 INDICATIONS

FOR SAECG

Prediction of Arrhythmic Events After MI Studies concerning patients after acute MI have shown that an abnormal SAECG does predict an increased likelihood of developing spontaneous VT or sudden cardiac death (TabIe 3). In these studies, an abnormal SAECG was present in 21% to 44% of patients recovering from an acute MI. Patients who had abnormal SAECGs developed sustained VT, VF, or sudden cardiac death with an incidence of 4%

AND

DENES

to 29% over approximately 1 year of follow-up, compared with a 0.8% to 4.3% arrhythmic event rate in patients who had normal SAECGs. The sensitivity of the SAECG in identifying those patients who are likely to develop arrhythmic events was 65% to 93%. Breidthardt et al found the SAECG to be more sensitive in identifying patients who developed sustained VT than those dying suddenly. h9 This is consistent with observations that late potentials are more common in patients presenting after spontaneous sustained VT than those presenting with primary VF. Several studies have compared the SAECG with other clinical parameters regarding the ability to predict arrhythmic events after MI. The SAECG has somewhat higher sensitivity (75%-93%) than parameters of depressed left ventricular function (67%-92%) or frequent or complex ectopy on Holter monitoring (50%80%).s7.7y-x2Cripps et al found that the clinical evidence for left ventricular dysfunction, Killip Class 2 II, was very specific (91%) but insensitive (55%) in identifying patients who were prone to arrhythmic events.xl Multivariate analysis using stepwise logistic regression has shown left ventricular ejection fraction 540% to be the most powerful independent predictor of arrhythmic events after MI followed by abnormal SAECG and complex ectopy on the Holter monitor, in that order.7y,x0 In a subsequent study, Gomes et al reported that SAECG had a higher predictive value for arrhythmic events than ejection fraction in patients who had acute anterior MI whereas these tests had equivalent predictive value after inferior infarction.8” El Sherif et al found the SAECG and depressed left ventricular function to be independent predictors of arrhythmic events, but complex ectopy on Holter monitoring was not.57 The independent nature of the variables of SAECG, left ventricular function, and Holter monitoring allows the use of combinations of these factors to predict high- and low-risk subgroups. The SAECG is particularly useful in predicting the risk for arrhythmic events in patients who have left ventricular dysfunction. Patients who have an ejection fraction <40% but who have a normal SAECG have a risk of arrhythmic events of 4% to 10%57~7y~8” which is similar to the risk of arrhythmic events in

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

Table

ECG

3. SAECG

339

to Predict

Arrhythmic

Events

After

MI

% Study

El-Sherifs7

Pt No

F/U MO

HPF U-W

156

12

25

SAECG Parameter

1 of2: RMS QRS

Breithardtc9

511

Subset

Abnormal SAECG

25 5 25 > 120

Visual

18

222

38

Arrhythmic Events #

Sells %

Spec %

-i-Pred %

-Pred %

AE

75

79

23

97

50 79

62 63

4 6

97 99

-

9

AE

SCD

4

VT VT-VF

3 5

LP > 10ms

SCD VT

16 14

SCD VT

Visual

VT

11

VT

LP > 10ms von

Leitner75

518

10

21

100

Denniss76

306

12

.05

Kuchar”

210

14

40

Games’*

102

12

40

Verzoniso

159

220

12

8

25

40

CD

57

80

7.3

SCD

50

79

3.6

SCD VT-VF

AE

65

77

19

96

Either: RMS-40 < 20 PV QRS > 120ms

39

SCD VT

AE

93

65

17

99

One of: RMS-40 < 20 I.LV LAS > 38 ms

44

SCD VT VT-VF

5 7

AE

87

63

2!9

96.5

24

SCD VT

5 6

AE

91

81

26

99

28

SCD VT

3 3

AE

83

73

8

99

of follow-up;

HPF,

> 140ms

3

> 114ms

1 of3: RMS < 25 LAS > 25 QRS

> 120

2of3: RMS

2 20

LAS t 40 QRS z 120 Abbreviations: ventricular fibrillation: value.

Pt No, activation time; AE, arrhythmic

number

of patients

98.5 99.1

26

VAT

QRS Cripps’9

CD SCD

in study;

F/U MO, months

LP, late potential; CD, cardiac death; SCD, events; Sew., sensitivity; Spec, specificity;

patients who have ejection fraction > 40% (5%8%). By contrast, patients who have both left ventricular dysfunction and an abnormal SAECG have a risk of 34% to 37% of developing arrhythmic events. The odds ratio for developing arrhythmic events with combined factors of left ventricular dysfunction and an abnormal SAECG was 30- to 39-fold, and it was 5.6- to 8.7-fold for the combination of left ventricular dysfunction and complex ectopy on Holter monitoring. Thus, these studies indicate that SAECG is superior to complex ectopy on 24-hour monitoring in determining risk for serious arrhythmic events after MI. High-risk populations (arrhythmic event rates of 31%-63%), defined by the presence of abnormal SAECG coupled with

high-pass

filter

sudden cardiac death; VT, ventricular +Pred, positive predictive value;

(corner

-frequency);

VAT,

tachycardia; VF, ventricular -Pred, negative predictive

either left ventricular dysfunction or complex ectopy on Holter monitoring are identifiable and may be appropriate targets for future interventions to prevent lethal arrhythmias after acute MI. Low-risk populations are defined by a normal SAECG coupled with normal left ventricular function or absence of complex ectopy on Holter with a risk of 0% to 1% for arrhythmic events.57~79~s0 of Nonsustained VT The use of SAECG in risk stratification of patients who have remote infarction and nonsustained VT appears promising. Several studies have shown that patients who have abnormal SAECG are more likely to have inducible susEvaluation

DUNBAR

340

tamed VT with programmed electrical stimulation.55J4-87 Recent studies of patients who have remote MI and nonsustained VT have shown that patients who have inducible sustained monomorphic VT are at higher risk for spontaneous VT, VF, or sudden cardiac deathR8-90than those patients who are noninducible and are at low risk for these eventsx8-91; and that electrophysiologically guided therapy may decrease the potential for lethal arrhythmias.88,90 Thus, SAECG could be applied as an inexpensive, noninvasive screening test to select those patients who have remote MI and nonsustained VT who could benefit from electrophysiologitally guided antiarrhythmic therapy. This is especially pertinent given the failure of ventricular premature beat suppression guided by Holter monitoring to prevent sudden cardiac death in patients after MI as shown in the Cardiac Arrhymia Suppression Trial.9’ Nalos et al evaluated the use of the SAECG as a screening test for inducible sustained VT in “high risk” patients presenting with syncope, nonsustained VT, VT, and sudden cardiac arrest.84 An abnormal SAECG had a sensitivity of 93% and a specificity of 94% in predicting inducible sustained VT in this population and was independent of left ventricular ejection fraction, presence of ventricular aneurysm, MI, and other clinical variables. Buxton et al evaluated 50 patients who had remote MI and nonsustained VT with SAECG and programmed electrical stimulation.55 In this study, SAECG abnormalities identified patients who had inducible sustained VT and inferior MI, but the study was not useful in identifying patients who had anterior MIS. A strategy for risk stratification and management of patients who have nonsustained VT and organic heart disease using evaluation with programmed electrical stimulation, the SAECG, and left ventricular function was recently described by Turrito et alR7 In this study, 90 patients who had nonsustained VT were evaluated. Sixty-three patients had underlying coronary artery disease and 27% had cardiomyopathy. Late potentials were present in 23 of 90 patients (26%). Late potentials identified 15 of 22 patients as having inducible VT, 68% sensitivity with a positive predictive value of 65%. The

AND

DENES

absence of late potentials had a high specificity (89%) and a negative predictive value of 90%. VT was inducible in 13 of 19 patients (68%) with a late potential and an ejection fraction <40%. Seven of 27 patients (21%) had inducible VT with depressed left ventricular function but no late potential, and two of four patients who had late potentials but good left ventricular function had inducible VT. The 3-year sudden death rate was 19% in the group with inducible sustained monomorphic VT and 8% in patients who did not have inducible tachycardia. There was no inducible VT in 33 patients who had normal SAECG and an ejection fraction > 40%. These patients had a low (3-year) sudden death rate of 7% without antiarrhythmic drug therapy. The investigators recommended a strategy of evaluating patients who had organic heart disease in nonsustained VT with programmed electrical stimulation only if the ejection fraction or the SAECG is abnormal. Unexplained Syncope Evaluation

The SAECG may provide additional information in patients presenting with unexplained syncope by identifying those who are more likely to have inducible VT during programmed electrical stimulation. Winters et al reported on 40 patients with syncope of unknown origin who underwent SAECG and programmed electrical stimulation.9-l Of the 34 patients without bundle branch block, 9 had inducible VT and 25 were noninducible. An abnormal SAECG parameter was present in 92% of patients with inducible VT, but was present only in 27% of those who were noninducible. In this study, RMS-40 voltage < 20 mV was the best discriminator with a sensitivity of 83% and a specificity of 92%. In a study by Gang et al of 24 patients who had unexplained syncope, filtered QRS of more than 120 milliseconds identified 8 of 9 patients who had inducible VT with a sensitivity of 89%.y4 None of 15 patients who were noninducible had a filtered QRS > 120 milliseconds (specificity and negative predictive value 100%). Kuchar et al also reported that SAECG had a high sensitivity of 73% and specificity of 89% in identifying patients with syncope attributed to VT in a study of 150 patients.95 During a median follow-up of 10 months, there were 15 deaths

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

(10%) of which six were sudden. There was no significant difference in recurrent syncope or survival between patients with or without late potentials. Vatterot et al reported that patients who had a normal SAECG had a 97% to 98% chance of no inducible VT in a series of 129 patients evaluated for syncope of uncertain origin.‘6 Although these studies suggest that the SAECG adds to the evaluation of patients with syncope of uncertain origin, its clinical utility has not been completely defined.‘” In patients who have unexplained syncope, electrophysiological studies may identify causes of syncope other than VT and VT may be inducible in certain patients who have normal SAECG. Evaluatiorl After Surgery For VT

The SAECG has been used to evaluate patients after surgical treatment for recurrent VT. EarIy reports by Uther et aIf and Rozanski et al”’ described abolition of late potentials after successful surgery for recurrent VT. Several studies have reported that VT was no longer inducible in 90% to 100% of patients in whom late potentials disappeared after endocardial resection for VT.97-‘00 VT was no longer inducible in 43% to 66% of patients in whom late potentials were still present after surgery in these series. Breithardt et al recently reported limited clinical utility of the SAECG in evaluating the success of surgery for recurrent VT.‘O’ In this series of 86 patients, 40 (46%) did not have late potentials preoperatively. Overall, the SAECG had a sensitivity of 36% and specificity of 69% in identifying patients who were noninducible postoperatively. The investigators believed that programmed electrical stimulation was mandatory in the individual patient to assure postoperative noninducibility. Other Indications for SAECG

Although one of the original clinical applications of SAECG was His bundle recording,],’ the SAECG has a rather limited role in the evaluation of atrioventricular conduction. Successful His bundle recordings can be obtained in SO% to 75% of patients. However, overlap of the His bundle potential by atria1 activity limits evaluation in up to 40% of patients.‘O” Documentation of the site of second degree atrioventricu-

341

lar block to distinguish supra-His block from infra-His block is limited with the current R wave triggered SAECG analysis. Takeda et al were able to distinguish supra-His from infraHis block by selective P wave triggering.“‘” Currently, few clinical decisions are made on the basis of HV interval alone. The SAECG has also been used to define electrophysiological mechanisms of tachycardias. Kuchar et al described the use of the SAECG to identify atria1 activation through analysis of P wave morphology and to determine the mechanism of supraventricular tachycardia by the QRS to P wave relationship. lo4Three patients who had atrioventricular reciprocal tachycardia were identified with ventricular-atria1 intervals exceeding 90 milliseconds. The two patients with left lateral bypass tracks had a left to right P wave axis on the unfiltered leads. Of the six patients who had atrioventricular nodal or reentrant tachycardia, three patients had atria1 activation within the QRS characterized by the absence of a P wave, and three patients had a P wave seen at the end of the QRS complex with a superior P wave axis. During atria1 flutter with 2:l AV conduction, F waves could be identified immediately before and after the QRS complex with a superior frontal plane F wave axis. Patients who have primary atria1 tachycardia tended to have variable ventricular response, limiting the R wave triggered signal-averaging process from readily identifying P waves. Thus, the SAECG could provide a noninvasive method for recording atria1 activation during supraventricular tachycardia and provide clues to the mechanisms of the tachycardia. A recent report described detection of patients who are at risk for atrial fibrillation (using P wave to trigger SAECG analysis) during sinus rhythm.“‘” In this study, the RMS voltage of the last 10 or 20 milliseconds of the P wave and the duration of the P wave were increased in patients who had paroxysmal atria1 fibrillation compared with control patients. The SAECG has been examined as a marker for cardiac rejection after orthotopic cardiac transplantation in man. Keren et al reported that the SAECG was more reproducible and more accurate than standard electrocardiography in detecting reduction of QRS voltage

342

DUNBAR

associated with definite cardiac rejection.‘Oh The method, however, was inadequate in monitoring for rejection in the early postoperative period and in detecting mild rejection in the last postoperative period. Haberl et al reported an increase in high-frequency spectral content in 19 of 20 patients who had acute rejection with fast Fourier transform analysis but no significant changes with time domain analysis during rejection in most patients.‘07 SAECG IN NONISCHEMIC

CARDIOMYOPATHY

Relatively little information is available regarding the role of SAECG and risk stratification and management of patients who have nonischemic cardiomyopathy. In the largest scries published to date, Poll et al reported on 41 patients who had nonischemic cardiomyopathy, 12 of whom had a history of sustained VT or VF.‘O* Patients who had a history of sustained ventricular arrhythmias had a significantly longer filtered QRS duration and lower RMS-40 voltage than those patients who did not have a history of sustained ventricular arrhythmias. Filtered QRS duration was prolonged in 92% and RMS voltage was abnormal in 83% of the patients who had a history of sustained ventricular arrhythmias compared with prolonged filtered QRS in 34% and low RMS voltage in 17% of those patients who did not have sustained arrhythmias. When both SAECG parameters were required for an abnormal test, the SAECG had a 83% sensitivity and an 86% specificity in identifying cardiomyopathy patients who had sustained VT or VF. Thus, this study indicated that the SAECG could identify those patients who had nonischemic cardiomyopathy and VT or VF. These conclusions are supported by other studies.‘0y-1’2 However, Denniss reported a lower incidence of late potentials in patients who have VT/VF with nonischemic cardiomyopathy than in those who have ischemic heart disease.“’ Coto et al found that late potentials could not accurately identify patients who have symptomatic nonsustained VT unrelated to an MI.“’ In this study, those patients who had cardiomegaly and symptomatic nonsustained VT had abnormal RMS-40 values whereas patients who had normal heart size and symptomatic nonsustained VT did not. Middlekauf et al

AND

DENES

recently reported a much lower incidence of late potentials (14%) in patients who had dilated cardiomyopathy evaluated for cardiac transplantation than in patients who had an old MI (40%).‘14 In this study, the presence of late potentials was a poor predictor of risk for sudden cardiac death, which occurred in 12% of patients who had late potentials and 21% of patients who did not have late potentials. Cripps et al recently reported evaluation of 64 patients who had hypertrophic cardiomyopathy with SAECG and Holter monitoring.ii5 An abnormal SAECG was present in 20% of patients who had hypertrophic cardiomyopathy compared with 4%1 of age- and gender-matched control patients. Abnormalities of the SAECG were associated with the presence of nonsustained VT on Holter monitoring. SAECG abnormalities were present in 8 of 17 patients who had nonsustained VT, 6 of 7 patients who had greater than three episodes of nonsustained VT per 24 hours, and 3 of 4 patients who had a history of cardiac arrest. An abnormal SAECG was present in only 5 of 47 patients who did not have nonsustained VT on Holter monitoring. When the SAECG was used to detect electrical instability in these patients (defined as a history of a cardiac arrest or the presence of nonsustained VT on Holter monitoring), it had a sensitivity of SO%, specificity of 93%, and a positive predictive accuracy of 77%. Of note, an abnormal SAECG was present in 2 of 3 patients less than 25 years old who had a cardiac arrest and was not present in the other 17 patients less than 25 years old who did not have a history of prior cardiac arrest. Gavaghan et al also noted an association between an abnormal SAECG and the presence of nonsustained VT on Holter monitoring in patients who have hypertrophic cardiomyopathy. ’ Ih Fananapazir et al have reported that SAECG is a useful test to predict which patients with hypertrophic cardiomyopathy will have inducible VT during programmed electrical stimulation.“’ The SAECG may be useful in evaluating patients who have VT arising from the right ventricle (VT with left-bundle-branch block [LBBB] configuration). Late potentials were first identified clinically in young patients who had VT due to arrhythmogenic right ventricular

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

dysp1asia.s Marcus et al found late potentials in 13 of 16 patients with arrhythmogenic right ventricular dysplasia who had SAECG recording. l’* Seven of these patients had ventricular postexcitation waves suspected on the resting 12-lead electrocardiogram. Similar results were reported by Bloomstram-Lindquist et al with late potentials being present in 16 of 18 patients who had right ventricular dysplasia.li9 In this study, filtered QRS duration was markedly prolonged (146 ? 19 ms) and was longer in the right compared with the left precordial leads. Hombach et al have reported a high incidence of late potentials in postoperative tetralogy of Fallot patients and found a correlation with subsequent occurrence of VT.r*O An example of an abnormal SAECG in a patient with arrhythmogenic right ventricular dysplasia is shown in Fig 3. By contrast, SAECG is normal in patients who have repetitive monomorphic VT, or VT arising from the right ventricular outflow tract with a LBBB and inferior axis configuration. Buxton et al reported that filtered QRS duration was normal in all 13 patients who had the form of right VT and that only 1 of 13 patients

Fig 3. Abnormal SAECG recorded from a l&year-old girl who had arrhythmogenic right ventricular dysplasia and symptomatic nonsustained VT. A late potential is present defined by abnormal RMS-40 amplitude of 8.67 PV and prolonged LAS 50.0 milliseconds. (Recorded with Corazonics Predictor 50 [Oklahoma City, OK) at corner frequency of 40 Hz. Courtesy of Dr. Steven Remole. University of Minnesota Hospitals.)

343

100

,

nn

Durations hs): Std QRS 86 Total QRS 99 ~As(4ouv) FUG Voltages

ItIsNoise

32

(uV‘I: 50 0.4

h

Fig 4. Normal SAECG recorded from a 58-year-old woman who had paroxysmal sustained VT with LBBB-type configuration and inferior frontal plane axis. This patient had a normal 12-lead electrocardiogram, normal coronary angiography, and no evidence of structural heart disease. (Recorded wlth Marquette with a corner frequency of 40 Hz. Courtesy of Dr. Martin Charlat, Minneapolis.)

had RMS-40 amplitude < 25 ~V.i*i Mehta et al reported on the relationship of SAECG in 38 patients who had VT unassociated with clinically apparent heart disease.122 In this series of 23 patients who had VT with LBBB and right axis deviation, 22 (91%) had normal SAECGs. These patients usually had normal endomyocardial biopsies and did not have inducible VT with programmed electrical stimulation. Figure 4 depicts a normal SAECG from a patient who had sustained paroxysmal VT with LBBB morphology and inferior frontal plane axis without underlying structural heart disease. Seven of 15 patients who had other morphologies of VT had abnormal SAECGs. These patients typically had abnormal endomyocardial biopsies and VT was inducible with programmed electrical stimulation. In this series, late potentials on the SAECG were insensitive for abnormal histologic findings (63%) or inducible VT (37%) but were very specific for these variables (84% and lOO%, respectively).

344

DUNBAR

AND DENES

REFERENCES 1. Berbari EJ, Lazzara R, Samet P, et al: Noninvasive technique for detection of electrical activity during the P-R segment. Circulation 48:1005-1013, 1973 2. Flowers NC, Horan LG: His bundle and bundlebranch recordings from the body surface. Circulation 48:IV102. 1973 (abstr) 3. Boineau JP, Cox JL: Slow ventricular activation in acute myocardial infarction: A source of re-entrant premature ventricular contractions. Circulation 48:702-713, 1973 4. Waldo AL, Kaiser GA: A study of ventricular arrhythmias associated with acute myocardial infarction in the canine heart. Circulation 47:1222-1228. 1973 5. El-Sherif N, Scherlag BJ, Lazzara R, et al: Re-entrant ventricular arrhythmias in the late myocardial infarction period: Conduction characteristics in the infarction zone. Circulation 55:686-702. 1977 6. Berbari EJ, Scherlag BJ, Hope RR, et al: Recording from the body surface of arrhythmogenic ventricular activity during the ST segment. Am J Cardiol41:697-702,197s 7. Josephson ME, Horowitz LN, Farshidi A: Continuous local electrical activity: A mechanism of recurrent ventricular tachycardia. Circulation 57:659-665. 1978 8. Fontaine G. Frank R, Gallais-Hamonno F, et al: Electrocardiography des potentials tardifs du syndrome de post-excitation. Arch Mal Coeur Vaiss 71:854-864, 1978 9. Breithardt G, Becker R, Seipel L, et al: Non-invasive detection of late potentials in man-A new marker for ventricular tachycardia. Eur Heart J 2:1-11, 1981 IO. Uther JB, Dennett CJ, Tan A: The detection of delayed activation signals of low amplitude in the vectorcardiogram of patients with recurrent ventricular tachycardia by signal averaging. in Sandoe E, Julian DG. Bell JW, (eds): Management of Ventricular Tachycardia-Role of Mexiletine. Amsterdam, the Netherlands, Excerpta Medica, 1978, pp SO-82 11. Simson MB: Use of signals in the terminal QRS complex to identify patients with ventricular tachycardia after myocardial infarction. Circulation 64:235-242, 1981 12. Breithardt G. Borggrefe M, Karbenn U, et al: Prevalence of late potentials in patients with and without ventricular tachycardia: Correlation with angiographic findings. Am J Cardiol49:1932-1937, 1982 13. Denes P. Santarelli P, Hauser RG, et al: Quantitative analysis of the high-frequency components of the terminal portion of the body surface QRS in normal subjects and in patients with ventricular tachycardia. Circulation 67:11291138.1983 14. Denniss AR, Richards DA, Cody DV, et al: Correlation between signal-averaged electrocardiogram and programmed stimulation in patients with and without spontaneous ventricular tachyarrhythmias. Am J Cardiol59:585-590. 1987 15. Denes P, Uretz E, Santarelli P: Determinants of arrhythmogenic ventricular activity detected on the body surface QRS in patients with coronary artery disease. Am J Cardiol53:1519-1523, 1984 16. Worley SJ, Mark DB, Smith WM. et al: Comparison of time domain and frequency domain variables from the

signal-averaged electrocardiogram: A multivariate analysis. J Am Coll Cardiol 11:1041-1051, 1988 17. Pollak SJ, Kertes PJ, Bredlau CE, et al: Influence of left ventricular function on signal averaged late potentials in patients with coronary artery disease with and without ventricular tachycardia. Am Heart J 110:747-752, 1985 18. Gomes JA, Winters SL, Stewart D, et al: Optimal bandpass filters for time-domain analysis of the signalaveraged electrocardiogram. Am J Cardiol 60:1290-1298, 1987 19. Kanovsky MS, Falcone RA, Dresden CA. et al: Identification of patients with ventricular tachycardia after myocardial infarction: Signal-averaged electrocardiogram, Holter-monitoring. and cardiac catheterization. Circulation 70:264-270, 1984 20. Buckingham TA. Ghosh S, Homan SM, et al: Independent value of signal-averaged electrocardiography and left ventricular function in identifying patients with sustained ventricular tachycardia with coronary artery disease. Am J Cardiol59:568-572. 1987 21. Gardner PI, Ursell PC, Fenoglio JJ, et al: Electrophysiologic and anatomic basis for fractionated electrograms recorded from healed myocardial infarcts. Circulation 72: 596-611.1985 22. Flowers NC, Wylas AC: Ventricular late potentials in normal subjects. Herz 13:160-168, 1988 23. Breithardt G, Cain ME, El-Sherif N. et al: Standard for analysis of ventricular late potentials using highresolution or signal-averaged electrocardiography. Circulation 83:1481-1488.1991 24. El-Sherif N, Restivo M, Craelius W. et al: The high-resolution electrocardiogram: Technical and basic aspects, in El-Sherif N, Samet P (eds): Cardiac Pacing and Electrophysiology (ed 3). Philadelphia, PA, Saunders. 1991, pp 349-371 25. Arthur RM: Fundamentals of time and frequencydomain analysis of signal-averaged electrocardiograms. Coronary Artery Disease 2:13-17. 1991 26. Atwood JE, Myers JJ, Forbes S, et al: High-frequency electrocardiography: An evaluation of lead placement and measurements. Am Heart J 116:733-739. 1988 27. Faugere G. Savard P, Nadeau RA, et al: Characterization of the spatial distribution of ventricular late potentials by body surface mapping in patients with ventricular tachycardia. Circulation 74:1323-1333, 1986 28. Berbari EJ, Ozinga L, Albert D: Methods for analyzing cardiac late potentials, in Ripley KL (ed): Computers in Cardiology, 1986. pp 35-40 29. Lander P, Berbari E: Use of high-pass tiltering to detect late potentials in the signal-averaged ECG. J Electrocardiol22:7-12. 1989 (suppl) 30. Cain ME. Ambos HD, Markham J, et al: Quantitation of differences in frequency content of signal-averaged ECGs in patients with compared to those without sustained ventricular tachycardia. Am J Cardiol55:1500-1505. 1985 31. Haberl R, Jilge G. Pulter R. et al: Comparison of frequency and time-domain analysis of the signal-averaged electrocardiogram in patients with ventricular tachycardia

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

and coronary artery disease: Methodologic validation and clinical relevance. J Am Co11 Cardiol 12:150-158, 1988 32. Caref EB, Turitto G, Ibramin BB. et al: Role of bandpass filters in optimizing the value of the signalaveraged electrocardiogram as a predictor of the results of programmed stimulation. Am J Cardiol64:16-26, 1989 33. Vatterot PJ, Bailey KR, Hammill SC: Improving the predictive ability of the signal-averaged electrocardiogram with a linear logistic model incorporating clinical variables. Circulation 81:797-804, 1990 34. Vacek JL, Smith S, Dunn MJ: Late potential parameter and noise variability caused by bandpass versus highpass filtering and type of signal averaging equipment used. J Electrophysiol3:278-285, 1989 35. Steinberg JS, Bigger JT Jr: Importance of the endpoint of noise reduction in analysis of the signal-averaged electrocardiogram. Am J Cardiol63:556-560, 1989 36. Vatterott P, Hammil S, Berbari E, et al: The effect of residual noise on the predictive accuracy of the signal averaged electrocardiogram. J Electrocardiol 2O:IV-102. 1987 (suppl) 37. Borbola J, Denes P: Short and long-term reproducibility of the signal-averaged electrocardiogram in patients with coronary artery disease. Am J Cardiol61:1123-1124, 1988 38. Oeff M. Leitner RV, Sthapit R: Methods for noninvasive detection of ventricular late potentials-A comparative multicenter study. Eur Heart J 7:25-33, 1986 39. Henkin R, Caref EB, Kelen G, et al: The signalaveraged electrocardiogram and late potentials. A comparative analysis of commercial devices. J Electrocardiol 2219, 1990 (suppl 1) 40. Moran JF, Larson L. Wilber D: Signal averaged electrocardiograms-A comparative study of two recording devices. IEEE Trans Biomed Eng 276:611-612. 1991 41. Poll DS, Marchlinski FE, Falcone RA, et al: Abnormal signal-averaged electrocardiograms in patients with nonischemic congestive cardiomyopathy: Relationship to sustained ventricular tachyarrhythmias. Circulation 72:13081313, 1985 42. Freedman RA, Gillis AM, Keren A. et al: Signalaveraged electrocardiographic late potentials in patients with ventricular fibrillation or ventricular tachycardia: Correlation with clinical arrhythmia and electrophysiologic study. Am J Cardiol55:1350-1353, 1985 43. Danford DA, Stelling JA, Kugler JD: Signal-averaged electrocardiography of the terminal QRS in healthy young adults. PACE Pacing Clin Electrophysiol 12:17121716, 1989 44. Simson MB, Untereker WJ, Spielman SR. et al: Relation between late potentials on the body surface and directly recorded fragmented electrograms in patients with ventricular tachycardia. Am J Cardiol51:105-112, 1983 45. Gann D, Balachandran PR, Sherif NE, et al: Prognostic significance of chronic versus acute bundle branch block in acute myocardial infarction. Chest 67:298-303, 1975 46. Hindman MC, Wagner GS, JaRo M. et al: The clinical significance of bundle branch block complicating acute myocardial infarction. Circulation 58679-688, 1978 47. Schneider JF, Thomas HE, McNamara PM, et al: Clinical-electrocardiographic correlates of newly acquired

345

left bundle branch block: The Framingham study. Am J Cardiol55:1332-1338, 1985 48. Freedman RA, Alderman EL, Shefheld LT. et al: Bundle branch block in patients with chronic coronary artery disease: Angiographic correlates and prognostic significance. J Am Coil Cardiol 10:73-80, 1987 49. Vatterott PJ, Bailey KR, Wiltgen CM, et al: The independent significance of QRS duration and ejection fraction in predicting arrhythmic events in asymptomatic complex ectopy patients. PACE Pacing Clin Electrophysiol 12:II-635, 1989 (abstr) 50. Fontaine JM, Henkin R. Howard M, et al: Establishing criteria for the presence of late potentials in patients with left bundle branch block. PACE Pacing Clin Electrophysiol lO:II-675, 1987 (abstr) 51. Buckingham TA, Thessen CC, Stevens LL, et al: Effect of conduction defects on the signal-averaged electrocardiographic determination of late potentials. Am J Cardiol61:1265-1271. 1988 52. Lindsay BD, Markham J, Schechtman KB. et al: Identification of patients with sustained ventricular tachycardia by frequency analysis of signal averaged electrocardiograms despite the presence of bundle branch block. Circulation 77:122-130, 1988 53. Breithardt G, Borggrefe M, et al: Prognostic significance of late ventricular potentials after myocardial infarction. Eur Heart J 4:487-495,1983 54. Gomes JA. Winters SL, Martinson M, et al: The prognostic significance of quantitative signal-averaged variables relative to clinical variables, site of myocardial infarction, ejection fraction, and ventricular premature beats: A prospective study. J Am Coll Cardiol 13:377-384, 1989 55. Buxton AE. Britton N, Simson MB: Application of the signal-averaged electrocardiogram in patients with nonsustained ventricular tachycardia after myocardial infarction: Implications for prediction of sudden cardiac death. J Electrocardiol21:S40-S45, 1988 (suppl) 56. McGuire M, Kuchar D, Ganis J, et al: Natural history of late potentials in the first ten days after acute myocardial infarction and relation to early ventricular arrhythmias. Am J Cardiol61:1187-1190,1988 57. El-Sherif N, Ursell S, Bekheit S, et al: Prognostic significance of the signal-averaged ECG depends on the time of recording in the post-infarction period. Am Heart J 118256-264.1989 58. Grimm M. Billhart RA, Mayerhafer KE. et al: Prognostic significance of signal-averaged ECGs during acute myocardial infarction: A preliminary report. J Electrophysiol21:283-288.1988 59. Hong M, Gang ES. Wang FZ, et al: Ventricular late potentials are associated with ventricular tachyarrhythmias in the early phase of myocardial infarction. Circulation 78:11-302,198s (abstr) 60. Eldar M, Leor J, Hod H, et al: Effect of thrombolysis on the evolution of late potentials within 10 days of infarction. Br Heart J 63:273-276, 1990 61. Gang ES, Lew AS, Hong M, et al: Decreased incidence ofventricular late potentials after successful thrombolytic therapy for acute myocardial infarction. N Engl J Med 321:712-716. 1989

346

62. Riccio C, Cesaro F, Perrotta R, et al: Early thrombolysis, reperfusion arrhythmias, and late potentials in acute myocardial infarction. New Frontiers of Arrhythmias 6:1.57161.1990 63. Lange RA. Agarro RG, Well PJ, et al: Incidence of antegrade flow in the infarct related artery on the incidence of late potentials after acute myocardial infarction. Am J Cardiol65:554-558. 1990 64. Simson MB, Falcone R, Dresden C, et al: The signal-averaged ECG and electrophysiologic studies in patients with ventricular tachycardia and fibrillation. Circulation 68:III-173, 1983 (abstr) 65. Borbola J, Ezri MD, Denes P: Correlation between the signal-averaged electrocardiogram and electrophysiologic study findings in patients with coronary artery disease and sustained ventricular tachycardia. Am Heart J 115:816824,198s 66. Dolack CL, Callahan DB, Bardy GH, et al: Signalaveraged electrocardiographic late potentials in resuscitated survivors of out-of-hospital ventricular fibrillation. Am J Cardiol65:1102-1104, 1990 67. Kertes PJ. Glaubus M, Murray A, et al: Delayed ventricular depolarization-Correlation with ventricular activation and relevance to ventricular fibrillation in acute myocardial infarction. Eur Heart J 5:974-983, 1984 68. Denniss AR, Holley LK, Cody DV, et al: Ventricular tachycardia and fibrillation: Differences in ventricular activation times and ventricular function. J Am Coll Cardiol 1:606, 1983 (abstr) 69. Breithardt G, Borggrefe M: Pathophysiological mechanism and clinical significance of ventricular late potentials. Eur Heart J 7:364-385, 1986 70. Simson MB, Kindwall E, Buxton AE, et al: Signal averaging of the ECG in the management of patients with ventricular tachycardia: Prediction of antiarrhythmic drug efficacy, in Brugada P, Wellens HJJ (eds): Cardiac Arrhythmias: Where to Go From Here? Mount Kisco, NY, Futura, 1987, pp 299-310 71. Simson MB, Spielman SR, Horowitz LN: Effects of antiarrhythmic drugs on body surface late potentials in patients with ventricular tachycardia. Am J Cardiol49:1030, 1982 (abstr) 72. Abdollah H, Brugada P, Richards DA: Influence of antiarrhythmic drugs on signal-averaged ECG. Circulation 68:111-422, 1983 (abstr) 73. Denniss AR, Ross DL, Richards DA, et al: Effect of antiarrhythmic drug therapy on delayed potentials detected by the signal-averaged electrocardiogram in patients with ventricular tachycardia after acute myocardial infarction. Am J Cardiol58:261-265,1986 74. Kuchar D, Thorburn C, Samuel N: Late potential duration predictsventricular tachycardia cycle length. Circulation 72:111-434. 1985 (abstr) 75. Borbola J, Denes P: Oral amiodarone loading therapy. I. The effect on serial SAECG recordings and the QTc in patients with ventricular tachycardia. Am Heart J 115: 1202.1208,1988 76. Simson MB, Falcone R, Kindwall E: The signal averaged electrocardiogram does not predict antiarrhythmic drug success. Circulation 72:7,1985 (suppl III) 77. von Leitner ER, Oeff M. Loock D, et al: Value of

DUNBAR

AND DENES

noninvasively detected delayed ventricular depolarizations to predict prognosis in post myocardial infarction patients. Circulation 68:111-83. 1983 (abstr) 78. Denniss AR, Richards DA, Cody DV, et al: Prognostic significance of ventricular tachycardia and hbrillation induced at programmed stimulation and delayed potentials detected on the signal-averaged electrocardiograms of survvors of acute myocardial infarction. Circulation 74:731-745. 1986 79. Kuchar DL, Thornburn CW, Sammel NL: Prediction of serious arrhythmic events after myocardial infarction: Signal-averaged electrocardiogram, Holter monitoring and radionuclide ventriculography. J Am Coil Cardiol 9531. 5381987 80. Gomes JA, Winters SL, Stewart D. et al: A new noninvasive index to predict sustained ventricular tachycardia and sudden death in the first year after myocardial infarction: Based on signal-averaged electrocardiogram. radionuclide ejection fraction and Holter-monitoring. J Am Coll Cardiol 10:349-357, 1987 81. Cripps T, Bennett ED, Camm HJ, et al: High gain signal averaged electrocardiogram combined with 24.hour monitoring in patients early after myocardial infarction for bedside prediction of arrhythmic events. Br Heart J 60:181187.1988 82. Verzoni A, Roman0 S, Pozzoni L: Prognostic signiticance and evolution of late ventricular potentials in the first year after myocardial infarction: A prospective study. PACE Pacing Clin Electrophysiol 12:41-51, 1989 83. Gomes JA. Winters SL, Martinson M. et al: The prognostic significance of quantitative signal-averaged variables relative to clinical variables. site of myocardial infarction. ejection fraction, and ventricular premature beats: A prospective study. J Am Coll Cardiol 13:377-384, 1989 84. Nalos PC, Gang ES, Mandel WJ, et al: The signalaveraged electrocardiogram as a screening test for inducibility of sustained ventricular tachycardia in high risk patients: A prospective study. J Am Coll Cardiol9:539-548, 1987 85. Winters SL, Stewart D, Targanski A, et al: Role of signal averaging of the surface QRS complex in selecting patients with nonsustained ventricular tachycardia and high grade ventricular arrhythmias for programmed ventricular stimulation. J Am Coll Cardiol 12:1481-1487. 1988 86. Lindsay BD, Ambos HD, Schechtman KB. et al: Improved selection of patients for programmed ventricular stimulation by frequency analysis of signal-averaged electrocardiogram. Circulation 73:675-683. 1986 87. Turrito G, Fontaine JM, Ursell S, et al: Risk stratification and management of patients with organic heart disease and nonsustained ventricular tachycardia: Role of programmed stimulation, left ventricular ejection fraction, and the signal-averaged electrocardiogram. Am J Med 88:1-37N, 1990 88. Gomes JA, Hariman RI, Kang PS, et al: Programmed electrical stimulation in patients with high-grade ventricular ectopy: Electrophysiologic findings and prognosis for survival. Circulation 70:43-51, 1984 89. Buxton AE, Marchlinski FE, Flores BT, et al: Nonsustained ventricular tachycardia in patients with coronary artery disease: Role of electrophysiologic study. Circulation 75:1178-1185, 1987

OPERATIONAL

ASPECTS

OF SIGNAL-AVERAGED

ECG

90. Klein RC, Machell C: Use of electrophysiologic testing in patients with nonsustained ventricular tachycardia: Prognostic and therapeutic implications. J Am Coil Cardiol 14:155-161, 1989 91. Wilber DJ. Olshansky B, Moran JF: Electrophysiologic testing and nonsustained ventricular tachycardia: Use and limitations in patients with coronary artery disease and impaired ventricular function. Circulation 82:350-358, 1990 92. The Cardiac Arrhythmia Suppression Trial (CAST) Investigators: Preliminary report: Effect of encainide and flecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infarction. N Engl J Med 321:406-412, 1989 93. Winters SL. Stewart D, Gomes JA: Signal averaging of the surface QRS complex predicts inducibility of ventricular tachycardia in patients with syncope of unknown origin: A prospective study. J Am Coil Cardiol 10:775-781, 1987 94. Gang ES. Peter T, Rosenthal ME, et al: Detection of late potentials on the surface electrocardiogram in unexplained syncope. Am J Cardiol58:1014-1020, 1986 95. Kuchar DL, Thornburn CW, Sammel NL: Signalaveraged electrocardiogram for evaluation of recurrent syncope. Am J Cardiol58:949-953, 1986 96. Vatterot PJ, Hammill SC, Bailey KR, et al: Signalaveraged electrocardiography: A new noninvasive test to identify patients at risk for ventricular arrhythmias. Mayo Clin Proc 63:931-942, 1988 97. Rozanski JJ. Mortara D, Myerburg RJ, et al: Body surface detection of delayed depolarizations in patients with recurrent ventricular tachycardia and left ventricular aneurysm. Circulation 63:1172-1178, 1981 98. Breithardt G, Seipel L, Ostermeyer J, et al: Effects of antiarrhythmic surgery on late ventricular potentials recorded by precordial signal averaging in patients with ventricular tachycardia. Am Heart J 104:996-1003, 1982 99. Marcus NH, Falcone RA, Harken AH, et al: Body surface late potentials: Effects of endocardial resection in patients with ventricular tachycardia. Circulation 70:632637, 1984 100. Denniss AR, Johnson DC, Richard DA, et al: Effect of excision of ventricular myocardium on delayed potentials detected by the signal-averaged electrocardiogram in patients with ventricular tachycardia. Am J Cardiol 59:591595,1987 101. Breithardt G, Borggrefe M, Martinez-Rubio A: Ventricular late potentials: Clinical aspects, in El-Sherif N, Samet P (eds): Cardiac Pacing and Electrophysiology (ed 3). Philadelphia, PA, Saunders, 1991, pp 384-400 102. McKenna WJ, Rowland E, Mortara D, et al: Noninvasive recording of the His bundle electrogram: Evaluation of the Marquette high resolution MAC unit, in Homback V, Hilger HH (eds): Signal Averaging Technique in Clinical Cardiology. Stuttgart, Germany, Schattauer, 1981, pp 301. 310 103. Takeda H, Kitamura K, Tsujimura T, et al: Noninvasive localization of A-V block by an “automated discrimination circuit.” in Homback V, Hilger HH (eds): Signal Averaging Technique in Clinical Cardiology, Stuttgart, Germany, Schattauer, 1981, pp 311-327 104. Kuchar DL, Kelley RP, Thorburn CW: High frequency analysis of the surface electrocardiograms of pa-

347

tients with supraventricular tachycardia: Accurate identification of atrial activation and determination of the mechanism of tachycardia. Circulation 74:1016-1026, 1986 105. Fukunami M, Yamada T, Ohmori M, et al: Detection of patients at risk for paroxysmal atrial fibrillation during sinus rhythm by P wave-triggered signal-averaged electrocardiogram. Circulation 83:162-169, 1991 106. Keren A, Gillis AM, Freedman RA, et al: Heart transplant rejection monitored by signal-averaged electrocardiography in patients receiving cyclosporine. Circulation 70:1-124-129, 1984 107. Haberl R, Weber M, Reichenspurner H, et al: Frequency analysis of the surface electrocardiogram for recognition of acute rejection after orthotopic cardiac transplantation in man. Circulation 76:101-108. 1987 108. Poll DS, Marchlinski FE, Falcone RA, et al: Abnormal signal-averaged electrocardiograms in patients with nonischemic congestive cardiomyopathy: Relationship to sustained ventricular arrhythmias. Circulation 72:130X1313,1985 109. Itoh S, Kobayashi K, Yoneda N, et al: Clinical study of late potentials: Comparison of late potentials in myocardial infarction, cardiomyopathy and idiopathic ventricular tachycardia. Jpn Circ J 52:21-29, 1988 110. Breithardt G, Borggrefe M, Karbenn U. et al: Prevalence of late potentials in patients with and without ventricular tachycardia: Correlation with angiographic findings. Am J Cardiol49:1932-1937, 1982 111. Denniss AR, Ross DL, Johnson DC. et al: Abnormalities on signal-averaged electrocardiogram and electrophysiologic study in patients with ventricular tachyarrhythmias not associated with ischemic heart disease. J Appl Cardiol 2:251-267, 1987 112. Ianucci G, Bacciarelle G, Villani M, et al: Late ventricular potentials in nonischemic congestive cardiomyopathy. PACE Pacing Clin Electrophysiol 10:690. 1987 113. Coto H, Maldonado C, Palakurthy P, et al: Late potentials in normal subjects and in patients with ventricular tachycardia unrelated to myocardial infarction. Am J Cardiol55:384-390, 1985 114. Middlekauff HR, Stevenson WG, Woo MA, et al: Comparison of frequency of late potentials in idiopathic dilated cardiomyopathy and ischemic cardiomyopathy with advanced congestive heart failure and their usefulness in predicting sudden death. Am J Cardio166: 1113- 1117, 1990 115. Cripps TR, Counihan PJ, Frenneaux MP, et al: Signal-averaged electrocardiography in hypertrophic cardiomyopathy. J Am Coil Cardiol 15:956-961. 1990 116. Gavaghan TP, Kelly RP, Kuchar DL: The prevalence of arrhythmias in hypertrophic cardiomyopathy: Role of ambulatory monitoring and signal-averaged electrocardiography. Aust N 2 J Med 16:666-670,1986 117. Fananapazir L, Barbour DJ. Winkler JB: Signalaveraged electrocardiographic identification of patients with inducible ventricular tachycardia with hypertrophic cardiomyopathy. Circulation 78:11-51, 1988 (abstr) 118. Marcus FI, Fontaine GH, Guiraudon G, et al: Right ventricular dysplasia: A report of 24 adult cases. Circulation 65384-398, 1982 119. Bloomstrom-Lindquist C, Hirsch I, Olsson B, et al: Quantitative analysis of the signal-averaged QRS in pa-

348

tients with arrhythmogenic right ventricular dysplasia. Eur Heart J 9:301-312, 1988 120. Hombach V: The high resolution electrocardiogram: Clinical aspects, in El-Sherif N, Samet, P (eds): Cardiac Pacing and Electrophysiology (ed 3). Philadelphia, PA, Saunders, 1991 pp 372-386 121. Buxton AE. Wanaman HL. Marchlinski FE, et al:

DUNBAR

AND

DENES

Right ventricular tachycardia: Clinical and electrophysiologic characteristics. Circulation 68:917-927, 1983 122. Mehta D, McKenna WJ. Ward DE, et al: Signihcance of signal-averaged electrocardiography in relation to endomyocardial biopsy and ventricular stimulation studies in patients with ventricular tachycardia without clinically apparent heart disease. J Am Coil Cardiol 14:372-379. 19X9