A Computerized approach to evaluating rest ... - Wiley Online Library

Clin. Cardiol. 5, 27-34 (1982) 0G. Witzstrock Publishing House, Inc.

A Computerized Approach to Evaluating Rest and Exercise-Induced ECG/VCG Changes after Cardiac Rehabilitation* K. WATANABE, M.D., v. BHARGAVA,Ph.D., v. FROELICHER. M.D.

Department of Medicine, Cardiovascular Division, University of California School of Medicine, San Diego, California USA

(Froelicher et al., 1980a). Though peripheral adaptations enable increases in functional capacity, only recently have central cardiac changes been suggested in selected patients (Atwood et al., 1980; Froelicher et al., 1980b; Jensen et al., 1980). However, there is no conclusive evidence for changes in ventricular function and myocardial proliferation in man. Demonstration of improvement in the electrocardiogram at rest and/or in response to the stress of exercise would be supporting evidence for cardiac rehabilitation. To fulfill the need for reproducible techniques to assess serial electrocardiographic studies, we have utilized a microprocessor assisted electrocardiograph/vectorcardiograph (ECG/VCG) digital acquisition system, refined recording techniques, and developed a spatial and scalar computer program for analyzing the ECG/VCG. This is a report of our methods of data acquisition and analysis for evaluating serial ECG/VCG changes. We also are reporting a search for possible electrocardiographic changes secondary to exercise training in coronary heart disease patients that could be measured in a larger, controlled study. To our knowledge, this is the first such study using computerized techniques to analyze both the resting and exercise 12-lead electrocardiogram and vectorcardiogram before and after an exercise training program.

Summary: A computerized method of acquiring and analyzing rest and exercise test 12-lead electrocardiographic and three-dimensional lead vectorcardiographic data before and after cardiac rehabilitations is described. Fourteen coronary heart disease patients were exercise tested before and after a mean of five months of aerobic exercise training. The only significant ST-segment improvements were found in three-dimensional space. Spatial measurements should be considered in the assessment of electrocardiographic changes secondary to exercise training.

Key words: cardiac rehabilitation, computerized exercise testing, coronary heart disease

Introduction Exercise training is recommended as a therapeutic modality for coronary heart disease (CHD) patients

* This work was supported by the Specialized Center of Research on Ischemic Heart Disease, NIH Research Grant HL 17682 from the National Heart, Lung and Blood Institute, Bethesda, Maryland, awarded to John Ross, Jr., M.D. Address for reprints:

Methods

Victor F. Froelicher, M.D. Director, Cardiac Rehabilitation and Exercise Testing University Hospital 225 Dickinson Street San Diego, CA 92 103

Fourteen patients with stable coronary heart disease are included in this study. Their age, resting electrocardiographic findings, and medications taken during this study are presented in Table I. Eleven patients were postmyocardial infarction and three had stable angina pectoris. The postmyocardial infarction patients were initially studied at least three months after their last

Received: February 26, 198 I Accepted with revision: September 25, 1981 21

Clin. Cardiol. Vol. 5, January 1982

28

TABLE1 Characteristics of the 14 patients with stable coronary heart disease included in this study ~~

Patients

~

Age

Resting ECG

I 2

51 58

3 4 5 6 7

54 60 60 64 60

8 9

57 54

10

31

II 12 13

54 43 49

14

41

LV H lnf Q wave Ant ST elevation Normal Normal Normal NSSTW changes Ant Q wave and S T elevation Normal I n f Q wave NSST W changes Ant Q wave and S T elevation Normal Normal Inf Q wave and Lat S T depression Inverted T waves in V3-V5

Drugs

Highest matched H R (Pre) (Post)

Months between tests while in exercise program

~

None lnderal

I23 I09

Ill

3

100

3

None None None Nitrates None

146 I24 I22 134 102

141 1 I5 I I4 137 I06

5 4 4 5 7

None lnderal

106 109

I06 96

7 4

Quinidine

I55

151

6

Dyazide None Nitrates

I30 I56 142

120 152 141

3 4 I

None

126

136

8

Abbreviations: LVH, left ventricular hypertrophy; Inf., inferior; Ant., anterior; Lat., lateral; NSSTW, nonspecific ST wave

acute myocardial infarction. None of these patients had serious complications such as congestive heart failure or serious dysrhythmias. They underwent supervised exercise training three times a week for at least three months. Dynamic exercise training included rowing, treadmill, arm ergometer, leg ergometer, steps, and rapid weight lifting. Before and after training, supine bicycle exercise testing was performed in order to evaluate the training effect. A multistage maximal supine bicycle exercise test was performed on an imaging table (Uniwork 845T, Quinton Instruments) because scintigraphic images were also obtained (Froelicher et al., 1980b; Jensen et al., 1980). The endpoint of testing was usually fatigue, though several patients were stopped due to angina pectoris and/or ST-segment depression of up to 0.3 mV. During exercise testing, blood pressure was taken at the end of each stage and continuous ECG/VCG data were recorded. ECG data acquisition was quality controlled by visualizing the recorded data on the strip chart recorder as it was entered on floppy disc. The data was digitized with 12 bit resolution at 250 samples/s. A MAC I data logger (Marquette Electronics) was used which utilizes a Motorola 6800 microprocessor as the central processing unit and a Motorola bus to communicate with the peripheral devices. Date, patient identification, and blood pressure were entered on the floppy discs by the digi-switches on the data logger. ECG and VCG data were collected using the Mason and Likar (1 966) exer-

cise adaptation of the 12-lead ECG and the Mayo Clinic (Berson et al., 1978) adaptation of the Frank VCG leads so that 14 electrodes were used to derive I5 leads. The Dalhousie square was used to assure consistent electrode placement (Rautaharju et al., 1976). Three leads were recorded and digitized simultaneously in time coherent fashion as follows: at rest 5 s each for I, 11, I l l , aVR, aVL, aVF, V1-6, and 10 s for X. Y, Z; during exercise and recovery the acquisition durations were doubled. The data were stored on soft sector, single-sided, double density floppy discs for off-line processing. Digital ECG data for the waveform analysis was processed by a Burroughs Computer using software developed at University of California, San Diego (Watanabe et ul., 1980) based on Vectan 11 (Golden, et ul., 1975) and the USAFSAM program (Kaiser el a l . , 1977). Before processing on the Burroughs Computer, the data stored on floppy discs was transferred to magnetic tape using a Sigma 3 computer. Digital data for each group of three leads transferred from floppy discs to magnetic tape was read and then absolute spatial vector velocity (ASVV) and coincidence function (CF) were calculated for the detection of the QRS complex.

K. Watanabe er al.: Computerized exercise test and cardiac rchabilitation

C F lessens the impact of noise on the estimation of an accurate fiducial point. The detection of the Q R S complex was performed by setting a threshold for ASVV and C F which was 30% of the maximal ASVV and 15% of the maximal CF. At the point just beyond both thresholds, a window was opened and the peak on each beat in ASVV was searched for by moving backward 20 ms and forward 80 ms. Then the point 90% of the peak value after the peak was used as the temporary fiducial point. The isoelectric line was estimated as the 16 ms segment of least activity within 20 ms and 120 ms of ASVV prior to the fiducial point. The mean value of this segment was computed in each lead. For the correction of baseline drift, the modified cubic spline mathematical technique (Meyer and Keiser, 1977) was applied for consecutive PR segments or nodes. A smooth ECG baseline was obtained by subtracting the estimated baseline drift from the uncorrected ECG. Q R S complexes with a prior R-R interval less than 80% of the mean R-R interval were eliminated in order to exclude premature beats. T o classify each beat, product-moment correlation coefficient (r) was used. The r was calculated for ASVV in a 200 ms region for a beat while the next beat was moved back and forth by a maximum of plus and minus 10 ins in steps of 2 ms from the temporary fiducial point. The point of maximum correlation was used to align the beats and initial the final fiducial point. The first beat was called template one. When the r between template one and the ASVV of the second beat was less than 0.9, the beat was classified as a new template or template 2 and so on. Usually the number of templates was limited to three and consecutive beats, except those discarded, were classified into a maximum of three templates. The QRS complex in each lead was aligned according to the fiducial points found in ASVV. The onset and offset of the P, QRS, and T waves were determined by algorithms applied to ASVV as previously described (Watanabe et a l . , 1980). Measurements were made on the scalar Frank leads, on lead groupings V4-6 and also on the Eigenleads derived from the Frank leads. The Eigenleads consisted of U (the narrow width), V (the length), and W (the thickness) of the Eigenloop. Spatial or Eigenplane derived leads were based on the concept of containment of maximum energy of the Q R S complex in a single plane. The maximum energy is concentrated at the peak amplitude; i.e., in about one-half to two-thirds of the Q R S interval. Usually more than 70 to 80% of the Q R S energy is contained in this middle portion of the Q R S complex and it predominantly defines the Eigenvectors and Eigenplane. I n each lead, ST-segment amplitude was measured at .I-junction ( Q R S end) and at 60 ms after J-junction (ST60). The ST-segment slopes were measured between J-junction and 60 ms after J-junction. In the most recent

29

TABLE I1 Means and standard deviations of some of the hemodynamic parameters measured pre and post the exercise program

HR

Supine rest Exercise

Pre Post Pre Post

(bpm)

SBP (mmHg)

DP

7 2 f 12 64 f 9 a 130 f 22 130 f 24

1 3 2 f I8 137 f 12 172 f 21 182 f 22

9500f2031 8860 f 1600 22.465 f 6145 23,890 f 6550

Abbreviations: HR, heart rate (bpm); SBP, systolic blood pressure; DP, diastolic pressure (SBP X HR) a p < 0.01

version of our program, ST integral, ST index, and time-normalized ST segment were also available. Spatial vector length was calculated as the amplitude of J x ~ . ~2+ 2 2 a t any point in time.

+

Data Analysis E C G / V C G data of pre- and posttraining were analyzed during supine rest, maximal exercise, the highest matched heart rate during exercise, and at three minutes recovery. Numerical values of only the measurements of interest a t rest and a t matched maximal heart rates pre- and posttraining are presented. Segments of the exercise ECG/VCG were classified by heart rate as follows: 70-90,91-110, 1 1 1-130, 131-1 50, and 151 Then we chose the ECG/VCG segments from pre- and posttraining within the same range of heart rate but at the highest range. These matched heart rates are shown in Table 11. Statistical analysis was performed utilizing an analysis of variance program available on a Tektronix 405 1 microcomputer.

+.

Results The 14 patients ranged in age from 37 to 64 years of age, with a mean of 52 years. They were in the UCSD Cardiac Rehabilitation Program for 3 to 8 months (mean of 5 months) and all exhibited a 10 to 20% increase in functional capacity. None of the patients were taking digitalis preparations. As shown in Table Ill, there was a significant decrease in resting heart rate as one would expect with training. However, a t rest or a t matched maximal exercise heart rates no change was found in systolic blood pressure or double product. As shown in Table IV, Q R S duration, S-wave amplitude, Q R S angles, or R-wave amplitude showed no significant change. Individual R-wave amplitude changes are shown in Fig. 1. As shown in Table V , no significant difference was found in the standard ST-segment depression and slope

Clin. Cardiol. Vol. 5, January 1982

30

TABLEI I I

Means and standard deviations of the QRS parameters measured pre- and post-training QRS d (ms)

Supine rest

Pre Post Pre Post

Exercise ~

~~

96.8 f 95.2 f 98.6 f 98.0f

13 I1 15

13

% Energy out of

RwX (mv)

RwEV (mv)

swx

Eigenplane

1.03 f 0.47 1.01 f 0.44 0.85 f 0.40 0. 87f0. 38

1.34 f 0.50 1.25 f 0.50 1.19 f 0.50 1.13f0.45

-0.09 f 0.08 -0.06 f 0.05 -0.2 I f 0.1 -0.20fO.l

3.1 f 3.2 3.1 f 2.6 3.0 f 2.5 4.2 f 3.6

Frontal plane QRS angle 28' 23' 37" 32'

f 17 f 22 f 18 f 14

Transverse plane QRS angle -29' -13'

-52' -35'

f 41 f 39 f 41 f 43

~~

Abbreviations: QRS d, QRS duration; RwX, R wave ii. Frank X lead; RwEV, R wave in longest axis of Eigenplane; SwX, S wave in Frank X lead TABLEIV Means and standard deviations of conventional ST depression and slope computerized measurements in three dimensions pre and post the exercise program.

Supine rest Exercise

Pre 0.0 f 0.04 Post 0.0 f 0.04 Pre -0.13f0.09 Post-0.13f0.1

0.1 f 0.6 0.02 f 0.04 0.0 f 0.45 -0.04 f 0.04 0.01 f 0.6 0.02 f 0.03 0.1 f 0.52 -0.03 f 0.03 0 . 0 f 0 . 7 -0.11 f O . l 0 . 5 5 4 ~1.0 0.01 fO.l 0.01f0.1 0 . 2 8 f 0 . 7 -0. 10f0. 12 0 . 5 9 f 1 . 3

-0.8 f 0.39 -0.6 f 0.39

-1.0f 1 -0.83~0.8

0.0 f 0.04 0.1 f 0.3 0.01 f 0.03 -0.1 f 0.4 0.2 f 0.5 -0.1 f 0.1 -0.08f0.08 0.1f0.5

Abbreviations: J-jct, mV from PR isoelectric line

measurements in the most sensitive lead, Vs, or in any of the X, Y , Z Frank leads that view the heart threedimensionally. As can be seen in Fig. 2, the mean re-

Supine rest Max. exercise Mat. max. ex. Recovery Pre Post Pre Post Pre Post Pre Post NS NS NS NS

mV 2.751

X Lead R-wave amplitude 2.251.75

/

-

sponse to the exercise test both pre- and posttraining was an abnormal ST-segment depression in X, Y, and Vs. ST-segment depression and R waves were of greater amplitude in V5 than in Frank X lead. We analyzed ST-segment amplitude a t 60 ms after QRS end (ST60) since Simoons (1977) has demonstrated it to be the optimal criteria for ischemia. Table VI and Fig. 3 show that no significant difference was found in the ST60 mean measurement pre- and posttraining in the X, Y,and Z leads. However, a statistically significant improvement in ST60 posttraining was found in the Eigenplane V lead. Using either absolute values or the mean calculations, there appeared to be less ST-segment displacement after the exercise program that was only apparent in the Eigenplane. Discussion

-------------

2.251 1.751250.75 -

t

ts

+--

-6

IiffSl

Eigen V Lead R-wave amplitude

FIG. 1 R-wave amplitude changes in Frank X lead and the Eigenplane.

The increase in functional capacity possible for most cardiac patients, even those with poor left ventricular function, is an important benefit of an exercise program (Lee et al., 1979). This could be due possibly only to changes in the peripheral response to exercise. However, improvement in myocardial perfusion and function secondary to chronic exercise has been demonstrated in animal studies (Froelicher et al., 1980a). Since the abnormal ST-segment shifts in coronary patients are most likely secondary to ischemia, lessening of such shifts would be consistent with improved myocardial perfusion.

K . Watanabe et al.: Computerized exercise test and cardiac rehabilitation TABLEV

31

Spatial ST-segment measurements (mV) made pre- and post-training

X ST60 Supine rest Exercise

Pre Post Pre Post

0.01 f 0.04 0.0 f 0.03

-0.08f0.12 -0.07f0.09

Y ST60

Z ST60

EV ST60

Absolute value EV ST60

SVL ST60

0.02 f 0.05 0.03 f 0.04 -0.08f0.13 -0.08f0.17

-0.09 f 0.05 -0.07 f 0.04 -0.05f0.13 -0.04f0.11

-0.03 f 0.08 -0.03 f 0.06 -0.12f0.12 -0.09"f0.09

0.07 f 0.05 0.05' f 0.04 0.13fO.l O.lO"f0.08

0.1 1 f 0.05 0.09' f 0.04 0.06 f 0.07 0.06f0.07

Abbreviations: ST60, S T amplitude 60 ms after QRS end (J-junction); EV, Eigenplane V lead; SVL, spatial vector length

' p < 0.01

For comparison purposes, only similar myocardial oxygen demands can be considered, therefore it is necessary to only compare the ST segments at matched exercise heart rate-systolic blood pressure pr,oducts. The product of heart rate times systolic blood pressure is the best noninvasive estimate of myocardial oxygen demand during exercise (Nelson et al., 1975). The following describes the previous investigations of the effect of an exercise training program on the exercise electrocardiogram. They are also summarized in Table VI. Costill and colleagues (1974) entered 24 men who demonstrated ST-segment depression in lead CM5 during treadmill testing into a 3-month exercise program. A second group of men with coronary heart dis-

ease and ST-segment depression were reexamined at the same interval, but were not trained. A third group of men with low physical fitness and no ST-segment depression were trained at the same relative intensity as the trained group with coronary disease. Maximal oxygen consumption was increased by approximately 25% in the trained groups. Training produced a lowering of heart rate for all submaximal exercise levels permitting the men to perform more work before the onset of angina and/or ST-segment depression (which occurred at the same heart rate before and after training). Training had no effect on the amount of ST-segment depression. Salzman and colleagues (1969) analyzed the exercise electrocardiograms of 100 men with coronary disease

Supine rest Max. exercise Mat rnax. ex. Recovery Pre Post Pre Post Pre Post Pre Post

rnv's[

NS

NS

NS

rnV

[

V, lead

2.001

-2.001

NS

Supine rest Max. exercise Mat rnax. ex. Recovery Pre Post Pre Post Pre Post R e Post NS NS NS NS

/

/ /

-------

\ L

2.000.00-

-2.00Z lead ST-OOms

2.001

/

Slope

/

//T -0.30

FIG.2 ST-segment slope and depression measurements in Vs, Y, and Z.

2 lead ST-J. amplitude

Clin. Cardiol. Vol. 5, January 1982

32 TABLEVI

Summary of previous pre- and post-training exercise test studies

Principal investigator (yr)

Number Number of trained controls

Length of exercise program

ECG lead monitored CM5 CdV, CH6, C4R

24 100

None

3 months 33 months

Kattus (1972)

13

15

5 months

CA5

Detry (1971)

14

None

3 months

CB5

Raffo (1 980)

12

12

6months

CM5

Costill (1974) Salzman (1969)

-

and a mean age of 48 years before and after an average of 33 months of cardiac rehabilitation. The exercise ECG improved in 3 1, did not change in 41, and deteriorated in 28 subjects. Improvement in the exercise ECG was more likely to occur in the group that showed improvement in physical fitness, while it was more likely to deteriorate in the group that showed a worsening of physical fitness. Improvement in the exercise ECG occurred in 80% of subjects with initial borderline or abnormal exercise ECG who had an improvement in physical fitness. Deterioration of the exercise ECG occurred in 70% of the patients with initially normal exercise ECG who had a worsening of physical fitness. In the subjects with an improved exercise ECG, the mean J-junction depression decreased from -0.17 to -0.08 mV and the mean slope increased from 0.16 to 0.91 mV/s. In the subjects with worsened exercise ECGs, the mean STsegment depression increased from -0.1 to -0.12 mV and the mean slope decreased from 1.0 to .25 mV/s. They concluded that improvement of the exercise ECG was related to enhancement of physiological function. Kattus and colleagues ( 1 972) identified 30 subjects with abnormal ST-segment depression in response to exercise testing without anginal pain in a screening study. All 30 of the abnormal responders were invited to enter a supervised exercise program. Two refused all further participation and fifteen, though unable to participate in the training program, agreed to return for a repeat treadmill test in the future. Thirteen of the abnormal responders participated in a supervised exercise program and increased their exercise capacity. Four of these had reversion of their ST-segment depression to normal (4/ 13,31%). Among those abnormal responders who did not train, there was no improvement of exercise capacity, yet two normalized their ECG patterns (2/15, 13%). Though not a randomized study, an untrained group was followed for comparison.

Description of subjects

Exercise ECG results

Three groups-see text MI, angina, and/or abnormal exercise ECG Asymptomatic with abnormal exercise ECG MI and/or angina

N o change in ST-segment response ST-segment changes correlated to changes in functional capacity Similar ST-segment improvement rate in controls

N o change in computerized ST-segment measurements at matched heart rate-systolic blood pressure product Angina with abnormal Higher heart rate for same exercise ECG amount of ST depression

Detry and Bruce (197 1) measured symptom limited maximal oxygen uptake and the ECG response to treadmill testing before and after three months of exercise training in 14 patients with coronary heart disease. ST-Segment responses were measured by computer averaging of 100 beat/samples. CB5 was recorded on magnetic tape during the last two minutes of each exercise level. ST measurements were made 50 to 70 ms after the nadir of the S wave. There was less ST-segment depression at submaximal exercise along with a lower double product. However, at maximal exercise both depression and double product were greater. The quantitative relationship of ST-segment depression to either exercise heart rate, the product of heart rate, and systolic blood pressure (double product) or to pressure rate product (heart rate times mean blood pressure) were unaffected by the exercise program. Maximal oxygen consumption increased by 21% and the product of heart rate and systolic blood pressure at symptom limited maximal exercise increased by 10% in the angina patients. Apparently, the only ECG changes were due to changes in the heart rate and blood pressure response to exercise rather than improved coronary circulation. Raffo and colleagues (1 980) studied 24 patients with stable angina and exercise-induced ST depression who were randomized into two groups. The 12 patients in group 1 followed the Canadian Air Force exercise program and the 12 patients in group 2 were controls. Exercise testing on a bicycle ergometer was performed at entry and six months later. Heart rate at the same level of ST-segment depression in CM5 and the duration of the test increased in those in the exercise program while heart rate at submaximal workloads decreased. Certain animal studies and echocardiographic studies have shown an increase in myocardial mass after chronic exercise (Froelicher et a l . , 1980a). Some studies have found an increase in R-wave amplitude in younger in-

K. Watanabe ef al.: Computerized exercise test and cardiac rehabilitation

mv[

0.301

-0’30L 0.801

Max. Supine rest exercise Mat m a . ex. Recovery Pre Post Pre Post Pre Post Pre Post NS NS NS NS

NS

P