Machler, Heinrich E. MD; Lueger, Andreas MD; Rehak, Peter PhD; Berger, Jutta MD; Veith, Wolfgang MD; Kuhbacher, Christoph MD; Koidl, Christoph MD; Stark, Gerhard MD; Metzler, Helfried MD
The concept of esophageal electrocardiography (E-ECG) is not new. In 1906, Cremer  passed an electrode into the esophagus of a professional sword swallower. Since then, numerous studies have confirmed the usefulness of this approach. Esophageal recording from a location close to the heart provides enhanced signal detection, and it has been used to diagnose dysrhythmias [2-4]. Until now, esophageal signals have been recorded with surface-ECG (S-ECG) amplifiers [5-7]. The literature does not report any E-ECG amplifier specifically designed to fulfill the particular requirements of esophageal recording.
Nevertheless, E-ECG recording has not become a routine clinical procedure. There is a single reported case in which developing myocardial ischemia or infarction was observed in the esophageal lead . No other data are currently available on the incidence of ischemia or electrical alternans detectable using S-ECG versus E-ECG or on measurements of ST segment alternans in the E-ECG.
Considering the limited sensitivity of intraoperative S-ECGs, the present study was a technical and methodological attempt to find a mode of analysis that would increase the sensitivity of a new, bipolar E-ECG. We report our initial experiences comparing S-ECG and high-resolution E-ECG, with special attention to the ST segment for early detection of myocardial ischemia.
Experiments were performed on 18 open-chest, anesthetized, and conditioned sheep of either sex after approval of our animal care committee. After being fasted overnight, animals weighing between 48 and 60 kg (mean 56 kg) were premedicated with xylazinehydrochloride (Rompun[registered sign]; Nicolet Instruments Corp., Madison, WI) 0.3-0.5 mg/kg body weight intramuscularly. Anesthesia was induced with ketaminehydrochloride 10% 10-20 mg/kg body weight and 5 mL etomidate intravenously. After tracheal intubation, anesthesia was maintained with small doses of isoflurane in combination with an infusion of fentanyl 0.3 micro g [center dot] kg-1 [center dot] min-1 and pancuronium 0.1 mg/kg. Positive pressure mechanical ventilation with air/O2 mixture maintained normoxia and normocarbia. Lactated Ringer's solution was infused at a rate of 3-5 mL [center dot] kg-1 [center dot] h-1 during the experiment. Taking into account the axis of the sheep heart and with knowledge of the isoelectric line , the electrodes for the leads of the S-ECG were implanted subcutaneously so as to be comparable to human leads. Each sheep had heparin-filled catheters inserted surgically after the induction of anesthesia via the right external jugular vein, and the arterial blood pressure was measured via the femoral artery. The hemodynamic variables were displayed on a monitor, recorded, and continuously interfaced to a computer.
The esophageal probe for E-ECG recording was inserted orally. A median sternotomy was performed, and the pericardium was opened over a length of 4-5 cm. Each sheep was allowed to recover for 2 h before examination. The body temperature was maintained constant with a heat lamp. The proximal diagonal branch of the left anterior descending coronary artery was identified and ligated with a monofil 6-0 suture to produce coronary occlusion. The chest was not closed. A video recording documented wall motion abnormalities and color changes of the ischemic area after coronary occlusion. At the end of the experiment, cardiac arrest was produced by potassium under deep anesthesia if spontaneous fibrillation did not occur.
The study was divided into three periods. The first period (steady state) covered recovery of the sheep from induction of anesthesia and sternotomy. In this period, S-ECG and E-ECG were monitored for 2 h. The second period (occlusion) followed the effects of occlusion during the first 2 min on both ECGs. The third period (acute ischemia and infarction) recorded both ECGs in sheep with acute myocardial ischemia or infarction during the next 30 min or until ventricular fibrillation occurred spontaneously.
After endotracheal intubation, the E-ECG probe was inserted transorally. The electrodes of the soft, flexible tube (diameter 7 mm) were positioned so that the E-ECG detected the maximal amplitude of the QRS complex. The distance was 70 cm from the edge of the alveolar ridge to the electrodes. The signals were recorded via a custom-made E-ECG lead containing three circumferential, chloridized silver electrodes (length 7 mm each) 25 mm apart . The proximal and the distal electrodes served as recording electrodes, and the intermediate electrode was the common reference electrode. We shielded the lead cables and twisted them together. Signals from the bipolar esophageal probe were amplified with a custom-made, battery-supported preamplifier (gain 1000, frequency range 0.01-2000 Hz, common mode rejection 140 dB, signal noise 5-7 micro V p-p). The preamplifier was galvanically separated from the rest of the equipment and was in accordance with the International Electrotechnical Commission standard for electronic appliances. A line frequency "notch" filter was not used. The preamplifier was developed in cooperation with the Institute of Medical Physics and Biophysics at our university. The E-ECG signal and the V5 signal were sent to a digital oscilloscope (Nicolet 3091[registered sign]) for simultaneous display and to a pulse code modulated (PCM)-recorder (Paar GmbH, Graz, Austria; 16-bit solution; sampling rate 47 kHz, frequency rate DC-15 kHz) for storage on video tape. Signals were simultaneously transmitted to an analog/digital board (Axotape[registered sign]; Axon Instruments, Inc., Foster City, CA; 12-bit solution, 2-kHz sampling rate) and stored on computer.
S-ECG data were recorded from the chest leads (V1-V5, V3 R-V5 R), as well as from the limb leads (I-III), monitored, and printed continuously on paper at 25 mm/s by a 12-lead ECG (Page Writer Xli[registered sign]; Hewlett-Packard, Palo Alto, CA; frequency range 0.05-150 Hz) to facilitate determination of ST segment elevation and ST segment depression for each minute of data acquisition.
The criteria used for detection of ischemia on S-ECG  were: 1) new ST depression (measured 60 ms after the J point) of >or=to1 mm in a horizontal or downsloping ST segment; 2) >or=to1.5-mm depression in a slowly upsloping ST segment; 3) ST segment elevation >or=to1.5 mm from baseline in a non-Q wave lead; or 4) ST elevation >or=to1.0 mm if an ST elevation of >or=to1.5 mm was present simultaneously in another lead. In the presence of significant resting J point elevation of >or=to 1 mm above the TP segment (early repolarization), ischemic changes were defined as a decrease to at least >or=to0.5 mm below the isoelectric baseline . Required duration of the change was >or=to 30 s. ST segment changes occurring during supraventricular or other dysrhythmias were not analyzed. A change in a single lead was also classified as an episode . Every S-ECG record was independently reviewed by two investigators (AL, GS) after the experiments.
During the three defined time periods (steady state, the first 2 min, and then the next 30 min, after the occlusion), the S-ECG and the E-ECG from each sheep were analyzed and compared.
Analyses of ST wave alternans were based on visual inspection of the E-ECG and on ST segment calculations. Therefore, 30 consecutive ST segments were visualized by means of a waterfall diagram to present time-dependent alterations. The same procedure was used for V5. The end of the T wave in the E-ECG was correlated with the end of the reference T wave in V5.
To calculate the dynamic changes of the ST segment, 200 beats were averaged using linking tracings with lines. The E-ECG was analyzed by determining the absolute difference in millivolts in the amplitudes of the ST segment between a mean of five successive beats (Amean5) and the next beat (AFB), performed for 200 consecutive beats at each millisecond of the ST segment (sum200 [vertical bar] Amean5 - AFB [vertical bar]). This means that we measured the differences between the mean value (Amean5) of the first to the fifth beat against the sixth beat (A (FB)), followed by measurements of the differences between the next Amean5 of the second to the sixth beat against the seventh beat (AFB), and so on for the next 200 beats. All recordings were done during normal sinus rhythm. Artifacts were excluded by visual inspection with interactive computer assistance.
Significant differences between ischemic S-ECG and ischemic E-ECG (Table 1) were calculated by using Fisher's exact test. The sum of the absolute differences in millivolts in the amplitude of 200 consecutive ST segments of all sheep before and during myocardial ischemia was calculated; a representative period (200th ms after the R peak) is presented in Table 2. The central tendency of the sum of the median values was calculated by the Hodges-Lehmann point estimator , and the 95% confidence intervals were calculated by using the Tukey test. A P value < 0.05 was considered statistically significant.
In all sheep studied, we found sinus rhythm at the induction of anesthesia and at steady state. Three sheep were excluded from the study; in two sheep, the esophageal recording was too noisy to analyze after the steady state, and one sheep developed ventricular fibrillation and died immediately after occlusion of the coronary artery. The waveform of the ECG and the E-ECG activity before myocardial ischemia are shown in Figure 1.
The recording in the E-ECG showed an early, low-amplitude part and a steep (prominent), high-amplitude biphasic P component, which occurred simultaneously with the P wave in the S-ECG during sinus rhythm. The mean amplitude of the biphasic P component in the E-ECG was 1.7 +/- 0.098 mV (range 1.1-2.6 mV), and the mean duration of the P components was 0.071 +/- 0.009 s (range 0.071-0.08 s). The mean voltage of the QRS segment was 3.9 +/- 3.5 mV (range 1.9-5.1 mV), the mean duration of the QRS segment 51 +/- 28.4 ms (range 28-89 ms), the mean voltage of the T wave was 1.4 +/- 1.7 mV (range 0.4-4.3 mV), and the duration of the T wave averaged 120 +/- 32.8 ms (range 76-160 ms).
During steady state, the S-ECG showed no signs of myocardial ischemia. The E-ECG presented with homogenous ST segments, without any beat-to-beat alternans (Figure 1).
In the second minute after occlusion, 7 of the 15 (47%) sheep revealed significant ST segment changes in the S-ECG (Table 1), whereas an increase in the ST alternans in the E-ECG (Table 1) was evident in 14 of the 15 sheep (93%) within 2 min of occlusion (P < 0.01). Electrical alternans was defined as a repetitive beat-to-beat fluctuation of E-ECG amplitude within the ST segment (Figure 2). All 15 sheep displayed wall motion abnormalities and a bluish discoloration in the ischemic area in this time period.
Distinct ST segment changes in the S-ECG were seen in 13 of the 15 animals (87%) in the interval from the 3rd to the 30th min. In the same observation period, all 15 sheep presented visible ST alternans on the E-ECG (Table 1, Figure 3). In the sheep with an increase in the ST alternans in the first 2 min, repolarization showed further exponential increase in nine cases in the next 30 min after the onset of ischemia.
The central finding was that the ischemic ST alternans was visually apparent in E-ECG but not in S-ECG. Using E-ECG, there was good correlation of the measurable ischemia (Table 2) with alternans to the morphology in the ST segment (Figure 2 and Figure 3). Coronary artery occlusion resulted in significant increases in the magnitude of the dynamic beat-to-beat alternation in ST segment amplitudes in the E-ECG signal (Figure 1, Figure 2 and Figure 3). The increase in ST segment alternans in the E-ECG significantly preceded the development of ST depression/elevation in the S-ECG (P < 0.01) (Table 1). We observed the development of "notched," "bifid," "dimpled," or "cloven" ST segments displaying humps, bulges, or protuberances before or just beyond the apex or on the descending limb of the T wave, above and below the isoelectric line (Figure 3). Frequently, the T wave alternans was concentrated during the first half of the T wave.
The main goals of this study were to evaluate the experimental esophageal recording of myocardial ischemia and infarction with a specifically designed preamplifier and to assess the association between ST segment alternans and ischemia and/or infarction.
ECG lead placement is important in detecting ischemia, especially intraoperatively and during cardiac surgery. Even with V5 monitoring, changes in lead position during chest opening and closure can lead to a failure to detect ongoing myocardial ischemia . Electromagnetic current dipole theory  predicts that reducing the distance between the recording electrodes and the cardiac signal source will result in a linear improvement in detection of subtle dynamic changes in the waveform. It must be assumed that the remote dipoles of surface leads would be less effective in detecting ST segment alternans than an esophageal electrode closer to the ischemic zone .
In the literature, there is only a single report of a patient with posterior ischemia who developed ST segment elevation only in an E-ECG and not in S-ECG leads II or V5 during a coronary bypass grafting procedure , and there has been no basic research on E-ECG and acute myocardial infarction .
The concept of measurement of ST segment alternans in the E-ECG is based on the idea that visible inhomogenities from beat-to-beat and increased differences in the amplitudes of the ST segment in the E-ECG lead reflect increased inhomogeneity in myocardial repolarization, providing a tool for tracking myocardial ischemia . The differentiation of electrocardiographic repolarization abnormalities indicative of acute myocardial processes from those that are the result of depolarization (QRS) alternans has been the subject of study, debate, and speculation for many years and is beyond the scope of discussion in this paper [17-19].
The lack of reports describing ST alternans recorded via the esophagus is surprising. Initially, E-ECGs were performed using S-ECG amplifier systems. Bipolar electrodes were simply connected to the terminals of a conventional S-ECG device [6,7], regardless of the low resistance and improved conduction properties of a bipolar esophageal lead. We connected our bipolar esophageal lead to a specifically designed preamplifier to increase the sensitivity .
The bipolar E-ECG presented herein was measured from the proximal electrode to the common electrode and against the distal electrode to the common electrode. This differs from a "pseudo" bipolar S-ECG, in which both electrodes are measured against the common electrode. Additionally, in S-ECGs, the reference electrode is more distant than it is with our E-ECG. This means that the E-ECG signals present a challenge that S-ECG equipment cannot meet.
Although an alternans can occasionally be identified on clinical ECG tracings, the alternans is usually so subtle than it cannot be detected visually . The S-ECG, with its specific filter technique, encounters high skin electrical impedance, low electrical conductivity, motion artifacts, potentials generated by the skin, and electromyographic noise. The baseline in the S-ECG may be stabilized by further filtering of low-frequency signals or by high-pass filtering . In contrast, the E-ECG is not contaminated by the wide range of voltages generated by physiological and other sources. With this in mind, we constructed a preamplifier with a wide frequency range and a bipolar lead.
During steady state, we observed very homogeneous ST segment intervals in the E-ECG (Figure 1). Immediately after the onset of ischemia, sheep presented visible ST alternans on the E-ECG. Repolarization alternans increased exponentially during the next 30 minutes after the onset of ischemia (Figure 2, Figure 3). This demonstrated the high sensitivity of the E-ECG, and the waterfall diagrams display consecutively inhomogenous repolarization periods (Figure 1, Figure 2 and Figure 3). Additionally, we calculated the sum of the differences of the mean net potentials and each next beat (Table 2). It is in good accordance with a report that the greater the variability in local repolarization processes, the greater the deviation of the sum of the net areas of QRS and T from zero . We propose that this calculation can serve as an index of heterogeneity of the repolarization process. Nearing et al.  showed no alternans in the nonischemic area in a canine model, and after occlusion, marked alternans of the T-waves, in good accordance with our results. They concluded that T wave alternans is an "electrophysiological characteristic of the ischemic myocardium" [18,19].
In humans, there remains discussion as to whether the E-ECG reflects more sensitively only the electrical potential of the posterior surface of the heart because of its close anatomical location to the posterior aspect of the left ventricle. When the axis of the sheep heart is compared with that of the human heart, the axis of the sheep heart is rotated to the right, and the relations between the anterior and posterior walls of the heart to the esophagus are fairly comparable.
Epicardial ECGs recorded simultaneously from the ventricle showed T wave alternans after left anterior descending coronary artery occlusion , in good accordance with the phenomenon in the E-ECG. We did not use epicardial electrodes because of the risk of injury and artifacts on insertion. Further work could involve different esophageal placements and different sizes of the electrodes, time series analysis, and Fourier transformations. The presented E-ECG signals were clearly seen and were easy to separate from artifacts, so we did not perform frequency analysis in both depolarizations and repolarizations. Additionally, further E-ECG studies should compare T wave alternans during the first and second halves of the T wave, and the "pilot" nature of these data require validation in humans. Clinical application and evaluation of the E-ECG to detect ST segment alternans could involve assessment of intraoperative myocardial ischemia, especially during cardiac surgery.
Based on our present findings, ST segment alternans in the E-ECG obtained from a bipolar lead positioned in the esophagus provided a remarkably high degree of sensitivity for the prediction of ischemia. Thus, quantification of the magnitude ST segment alternans may provide a useful tool for assessing vulnerability in cases of ischemia and repolarization disturbances. Despite the absence of visible ST segment alternans on the S-ECG, the E-ECG reveals a clearly definable peak of repolarization alternans with myocardial ischemia. The quantification of E-ECG ST segment is a promising technique for intraoperative ischemia detection compared with currently available methods.
1. Cremer M. Uber die direkte Ableitung der Aktionsstrome des menschlichen Herzens vom Osophagus und uber das EKG des Foten. Munch Med Wochenschr 1906;53:27-34.
2. El Sheruf N, El-Ramly Z, Sorour AH. Esophageal electrocardiography in the study of cardiac arrhythmias. Br Heart J 1969;31:414-7.
3. Jenkins JM, Wu D, Arzbaecher RC. Computer diagnosis of supraventricular and ventricular arrhythmias. Circulation 1979;60:977-85.
4. Schnittger I, Rodriguez IM, Winkle RA. Esophageal electrocardiography: a new technology revives an old technique. Am J Cardiol 1986;57:604-11.
5. Atlee JL. Perioperative cardiac dysrhythmias. 2nd ed. Chicago: Year Book Medical Publishers; 1990:150.
6. Atlee JL. Arrhythmias and pacemakers: practical management for anesthesia and critical care medicine. Philadelphia: WB Saunders, 1996:130-3.
7. Thys DM, Narang JN. Electrocardiographic monitoring in cardiac anesthesia: principles and clinical practice. In: Estafanous FG, Barash PG, Reves JG, eds. Philadelphia: JB Lippincott, 1994:93.
8. Kates RA, Zaidan JR, Kaplan JA. Esophageal lead for intraoperative electrocardiographic monitoring. Anesth Analg 1982;61:781-5.
9. Sporri H. Uber die Ausbreitung der Erregungswelle im Pferdeund Rinderherzen. Helv Physiol Acta 1956;C:79-81.
10. Machler HE, Lueger A, Bergmann P, et al. A high resolution esophageal ECG for monitoring atrial activity in the hypothermic, potassium-arrested heart. Anesth Analg 1997;84:484-90.
11. London MJ, Hollenberg M, Wong MG, et al. Intraoperative myocardial ischemia: localisation by continuous 12-lead electrocardiography. Anesthesiology 1988;69:232-41.
12. Hollander M, Wolfe DA. Non parametric statistical methods. New York: Wiley & Sons, 1973:123.
13. Zaidan JR. Initiation and maintenance of cardiac pulmonary bypass. In: Mora CT, ed. Cardiopulmonary bypass: principles and techniques of extracorporal circulation. New York: Springer-Verlag, 1995:266-7.
14. Gulrajani RM. Models of the electrical activity of the heart and computer simulation of the electrocardiogram. CRC Crit Rev Biomed Eng 1988;16:1-66.
15. Jain U, Rao TLK, Dasari M. Suitability of esophageal electrocardiogram for intraoperative monitoring [abstract]. Anesthesiology 1989;71:A342.
16. Verrier RV, Nearing BD. T wave alternans as a Harbinger of ischemia induced sudden cardiac death. In: Zipes DP, Jalife J, eds. Cardiac electrophysiology: from cell to bedside. 2nd ed. Philadelphia: WB Saunders, 1995:468.
17. Abbound S, Smith JM, Shargorodsky B, et al. High frequency electrocardiography of three orthogonal leads in dogs during a coronary artery occlusion. Pac Clin Electrophysiol 1989;12:574-81.
18. Fisch C. Electrocardiography and vectorcardiography. In: Braunwald E, ed. Heart disease: a textbook of cardiovascular medicine. 4th ed. Philadelphia: WB Saunders, 1992:116-60.
19. Surawitz B. ST-T abnormalities. In: Mac Faralne PW, Veitch Lawrie TD, eds. Comprehensive electrocardiography: theory and practice in health and disease. New York: Pergamon, 1989:511-63.
20. Rosenbaum DS, He B, Cohen RJ. New approaches for evaluating cardiac electrical activity: repolarization alternans and body surface Laplacian imaging. In: Zipes DP, Jalife J, eds. Cardiac electrophysiology: from cell to bedside. 2nd ed. Philadelphia: WB Saunders, 1995:1187-9.
21. London MJ, Kaplan JA. Advances in electrocardiographic monitoring in cardiac anesthesia. In: Kaplan JA, ed. Cardiac anesthesia. 3rd ed. Philadelphia: WB Saunders; 1993:306.
22. Castellanos A, Myergurg RJ. Primary and Secondary repolarization changes: the intraventricular conduction alteration model. ACC Current Journal review. 1995; January/February: 60-63.
23. Nearing BD, Huang AH, Verrier RL. Dynamic tracking of cardiac vulnerability by complex demodulation of the T wave. Science. 1991;252:437-440.
24. Nearing BD, Oesterle SN, Verrier RL. Quantification of ischemia induced vulnerability by precordial T wave alternans analysis in dog and human. Cardiovasc Res 1994;28:1440-9.