The Feasibility of Transesophageal Echocardiograph-Guided Right and Left Ventricular Oximetry in Hemodynamically Stable Patients Undergoing Coronary Artery Bypass Grafting

Abstract

There are no techniques available for continuous noninvasive measurement of the oxygen saturation of blood flowing through the heart. We assessed the feasibility and accuracy of transesophageal echocardiograph (TEE)-guided left ventricular (Spo₂ LV) and right ventricular (Spo₂ RV) oximetry. Twenty hemodynamically stable, well-oxygenated anesthetized patients (ASA physical status III, aged 51–75 yr) undergoing coronary artery bypass grafting were studied. A TEE probe was modified by attaching a single-use pediatric reflectance pulse oximeter just proximal to the ultrasound transducer. The TEE probe was directed toward the LV by using the transgastric mid-short axis view or toward the RV by using the transgastric RV inflow view, in random order. Readings were taken every 30 s for 10 min during a hemodynamically stable period of anesthesia. Simultaneous blood samples were taken from the radial artery and pulmonary artery to determine arterial oxygen saturation (Sao₂) and mixed venous oxygen saturation (Svo₂), respectively. During Spo₂ LV readings, simultaneous finger pulse oximetry (Spo₂ finger) was also recorded. Spo₂ LV was feasible in 20 of 20 patients, and Spo₂ RV was feasible in 19 of 20 patients. The mean ± sd (range) oxygen saturation for each method was the following: Spo₂ LV, 98.7% ± 0.6% (97%–100%); Sao₂, 98.7% ± 0.6% (96.6%–99.4%); Spo₂ finger, 98.1% ± 1.2% (97%–100%); Spo₂ RV, 73.9% ± 4.7% (64%–85%); and Svo₂, 74.5% ± 4.4% (66.8%–82.6%). Spo₂ LV agreed closely with Sao₂ (mean difference, 0.072%). Spo₂ RV agreed closely with Svo₂ (mean difference, 0.65%). Spo₂ LV agreed more closely with Sao₂ than finger oximetry (mean difference, −0.072 vs −0.692). TEE-guided Spo₂ LV and Spo₂ RV are feasible in hemodynamically stable anesthetized patients and provide similar readings to arterial and mixed venous blood samples. The technique merits further investigation.

Pulse oximetry is widely used in anesthesia and critical care medicine to provide noninvasive information about arterial oxygen saturation (Sao₂). Oximetry probes are usually applied to the fingers or ears for convenience, but this has a signal acquisition failure rate of 1.12% to 2.5%^(1–3), and more accurate readings can be obtained from better-perfused superficial tissues, such as the cheek ⁽⁴⁾, nasal septum ⁽⁵⁾, and tongue ⁽⁶⁾. More recently, oximeters have been placed into deep, vessel-rich areas, such as the esophagus ⁽⁷⁾, ^1,2 pharynx ^(8,9), and trachea ⁽¹⁰⁾, where they seem to provide more accurate readings than surface oximetry, even in hypoperfusion states. The tissue being sampled from these vessel-rich areas was assumed to be the surrounding mucosa with esophageal oximetry ⁽⁷⁾, but we recently found evidence that some of the signal is derived from deeper tissues, such as underlying large vessels, with pharyngeal or tracheal oximetry ^(8,10). The heart lies in close proximity to the esophagus, and we considered that esophageal oximetry readings might be partially derived from the heart. Furthermore, we considered that an appropriately located and directed esophageal oximetry probe might be able to derive oximetry readings from specific locations within the heart. Transesophageal echocardiography (TEE) is a relatively noninvasive technique for measuring cardiac performance during surgery and involves passage of an endoscope into the esophagus and ultrasound imaging technology ⁽¹¹⁾. In this study we assessed the feasibility and accuracy of TEE-guided left ventricular (Spo₂ LV) and right ventricular (Spo₂ RV) oximetry.

Methods

With research and ethical committee approval and written, informed consent, we studied 20 patients (ASA physical status III) scheduled for elective coronary artery bypass grafting. Patients were excluded if they had known esophageal or gastric pathology. A radial arterial catheter was inserted under local anesthesia into the nondominant hand, and a digital pulse oximeter was attached to the index finger of the dominant hand. Anesthesia was induced with midazolam 0.15 mg/kg, fentanyl 7 μg/kg, and rocuronium bromide 0.6 mg/kg and was maintained with oxygen 30% in air and a continuous infusion of fentanyl at 0.4 μg · kg⁻¹ · min⁻¹ and midazolam at 4 μg · kg⁻¹ · min⁻¹. A pulmonary artery catheter (right internal jugular vein) and the TEE probe (Omniplan HP 21364A; Ultrasound System; HP Image Point; Agilent Technologies GmbH, Vienna, Austria) were inserted. The TEE probe was modified by attaching a single-use pediatric reflectance pulse oximeter (Datex Medical Instrumentation, Helsinki, Finland) just proximal to the ultrasound transducer with an adhesive dressing 0.45 mm thick (Tegaderm; 3M, Ontario, Canada) that was placed up to the edge of, but did not cover, the optical components (Fig. 1). The position, orientation, and accuracy of the sensor were checked before and after use in each patient. The oximeter probes and monitors (Datex AS/3; Datex Medical Instrumentation) were identical for the TEE and finger. The monitors were checked to determine if they gave the same reading when attached to the same probe. The TEE probe was randomly placed in the following two positions: transgastric mid-short axis view (LV) or transgastric RV inflow view (RV).

Figure 1: Reflectance pulse oximeter attached to the transesophageal echocardiograph probe.

The transgastric RV inflow view was developed from the mid-short axis view by turning the TEE probe to the right until the RV cavity was located in the center of the display and by rotating the multiplane angle forward to between 90° and 120° until the apex of the RV appeared in the left side of the display. Rotating the multiplane angle forward to between 90° and 120° was used only to verify the correct position of the probe, not to change the position of the pulse oximeter. Once positioned correctly, the TEE probe was not adjusted until readings were complete. Spo₂ LV, Spo₂ RV, and finger Spo₂ were each measured over a 10-min period in random order during a hemodynamically stable period of anesthesia before the commencement of surgery. Oximetry data were collected every 30 s, giving a total of 20 data readings per patient for each form of oximetry. The average of these readings was used for each patient. Sao₂ or mixed venous blood (Svo₂) was collected via a vacuum container at the midpoint of each 10-min data collection sequence and was analyzed within 30 s of being drawn. The blood gas machine (Ciba Corning Series 800; Chiron Diagnostics GmbH, Salzburg, Austria) was accurate to 0.1% (Sao₂) and was calibrated before each case and used for cooximetry. Heart rate, mean blood pressure (arterial catheter), and percentage of inspired oxygen were recorded every 2 min during each 10-min data collection sequence. Patients were questioned about sore throat, dysphagia, and dysarthria 48 h postoperatively. Core (pulmonary artery catheter) and skin (hand) temperature were recorded at the midpoint of each data collection sequence.

TEE was performed by one anesthesiologist, data were collected by a second anesthesiologist, and Sao₂ and Svo₂ samples were collected by a technician. The first anesthesiologist performing the TEE was unaware of the readings, but the second anesthesiologist and the technician were aware of the type of oximetry being performed. Sample size (n = 20) was based on data from a pilot study with six patients (reporting a difference of the mean of 1 and an sd of 0.8) for a Type I error of 0.05 and a power of 0.9. The distribution of data was determined with Kolmogorov-Smirnov analysis ⁽¹²⁾. The levels of measurement agreement between TEE and finger sensors, as well as between Svo₂ and Sao₂ and each type of the probe, were calculated with the method outlined by Bland and Altman ⁽¹³⁾. The mean difference represents the average difference between each of the data points. Bland and Altman analysis is intended for analysis of two methods, neither of which is the “gold standard.” We have used Bland and Altman analysis for analysis of a new method and a “gold standard” method. The sem difference was calculated by dividing the sd by √n, where n is the sample size. The limit of agreement represents the mean difference ± 2 s (s = sd of the differences). The se of the limits of agreement was calculated by using the formula √(3 s²/ n).

Results

The mean (range) age, height, and weight were 66 yr (51–75 yr), 169 cm (154–189 cm), and 72 kg (52–101 kg), respectively. The male/female ratio was 14:6. The mean (range) core and finger temperatures during the readings were 36.8°C (36.4°C–37.1°C) and 36.7°C (36.3°C–36.9°C), respectively. The room temperature was 21°C. The mean ± sd (range) for heart rate was 79 ± 16 bpm (59–110 bpm); mean blood pressure was 81–12 mm Hg (60–112 mm Hg); and the inspired oxygen during readings was 31% ± 2% (28%–35%). The temperature and hemodynamic variables were stable during the data collection period. The mean (range) discrepancy between the two monitors was 0.1% (0%–2%). The position, orientation, and accuracy of the sensor were unchanged after use in each patient. Bland and Altman graphs for Sao₂ versus Spo₂ finger, Sao₂ versus TEE Spo₂ LV, and Svo₂ versus TEE Spo₂ LV are presented in Figures 2, 3, and 4, respectively. An example of the wave form from the LV, the RV, and finger is provided in Figure 5. The data set was 93% complete: Spo₂ LV, 394 (99%) of 400; Sao₂, 391 (98%) of 400; Spo₂ finger, 396 (99%) of 400; Spo₂ RV, 338 (85%) of 400; and Svo₂, 334 (84%) of 400. Spo₂ LV was feasible in 20 of 20 patients, and Spo₂ RV was feasible in 19 of 20 patients. In one patient, Spo₂ RV could not be obtained because of technical difficulties. The mean ± sd (range) oxygen saturation for each method was the following: Spo₂ LV, 98.7% ± 0.6% (97%–100%); Sao₂, 98.7% ± 0.6% (96.6%–99.4%); Spo₂ finger, 98.1% ± 1.2% (97%–100%); Spo₂ RV, 73.9% ± 4.7% (64%–85%); and Svo₂, 74.5% ± 4.4% (66.8%–82.6%). Between-method statistical comparisons for the oxygen saturation measurements are presented in Table 1. There were no complaints of dysphagia or dysarthria, but two patients had a mild sore throat.

Figure 2: Bland and Altman graph comparing the difference between arterial oxygen saturation from a radial arterial blood sample (Sao₂) and oxygen saturation from a finger pulse oximeter (Spo₂ finger) versus the mean oxygen saturation by the two methods

Figure 3: Bland and Altman graph comparing the difference between arterial oxygen saturation from the radial arterial blood line (Sao₂; current “gold standard” method) and oxygen saturation from the left ventricle (transesophageal echocardiography [TEE] Spo₂ LV; new method) versus the mean oxygen saturation by the “gold standard” and new method.

Figure 4: Bland and Altman graph comparing the difference between mixed venous oxygen saturation from the pulmonary artery catheter (Svo₂; current “gold standard” method) and oxygen saturation from the left ventricle (transesophageal echocardiography [TEE] Spo₂ RV; new method) versus the mean oxygen saturation by the two methods

Figure 5: An example of the wave form from the left ventricle, the right ventricle, and the finger.

Table 1:  Between-Method Statistical Comparisons for the Oxygen Saturation Measurements

Discussion

We found that TEE-guided LV and RV oximetry is feasible and accurate. Readings from the LV were <0.1% different from arterial samples, and readings from the RV were <1% different from SV samples. Pulse oximetry works by measuring the change in light absorption during blood flow in a constant mass of tissue. In finger, ear, nose, cheek, and tongue oximetry, the light emitter and sensor are aligned opposite each other, but in esophageal, pharyngeal, tracheal, and ventricular oximetry, they lie side to side. The high-quality signal we found with the side-to-side arrangement, as found in other studies ^(7,8,10), suggests that much of the emitted light is reflected off adjacent tissues and onto the sensor. The readings for Spo₂ LV were higher than Spo₂ finger and almost certainly reflect the greater perfusion of the LV and adjacent structures than the finger. Although we have not proven that the oximetry readings were coming from the RV and LV, it is likely that the majority of the signal was coming from these structures because the probes were directed at these chambers and the readings correlated closely with the blood samples. The wave forms were also different between the two chambers: the LV trace was similar to digital pulse oximetry with a single hump; the RV trace was wider and had a double hump (Fig. 5). The wave forms also remained constant over time. Furthermore, prestudy testing of the system on six patients showed that if the probe was directed toward the other ventricle, the readings became mixed.

During measurements of pharyngeal ⁽⁸⁾ and tracheal ⁽¹⁰⁾ oximetry in earlier studies with the laryngeal mask airway and tracheal tube, respectively, we found that the oximetry signal did not diminish when cuff pressures exceeded mucosal perfusion pressures. This suggested that light from the oximetry probe penetrates beyond the mucosal layer. We speculated that the pharyngeal oximetry was partially derived from the adjacent carotid artery and that left tracheal oximetry was partially derived from the adjacent left common carotid, the arch of the aorta, or both. This speculation was supported by data from an adult patient undergoing coronary bypass grafting in whom we found that tracheal Sao₂ precisely matched Sao₂ when the probe faced the left side and precisely matched Svo₂ when the probe faced anteriorly ⁽¹⁰⁾. Although it could be argued that the LV readings include, or solely comprise, tissue from the intervening myocardium, the RV readings so closely match those of RV blood samples that it is unlikely that they include any arterial contamination. This strongly supports the hypothesis that oximetry from these vessel-rich locations is not primarily mucosal in origin.

Svo₂ readings are often measured in critically ill patients to provide information about oxygen uptake, but these measurements require insertion of a pulmonary artery catheter with its attendant risks, and the readings are not continuous ⁽¹⁴⁾. This study shows that noninvasive continuous Svo₂ monitoring is feasible. The further development of this system might have applications in anesthesia, critical care medicine, cardiology, and pediatrics.

Our study has a number of limitations. First, we did not assess the accuracy of LV and RV oximetry over time, during transient changes, in low perfusion states, or during patient movement or swallowing. However, there is anecdotal ⁽⁹⁾ evidence that pharyngeal oximetry is accurate in low-perfusion states, and there is both anecdotal ⁽¹⁵⁾ and prospective ⁽⁷⁾ evidence ³ that esophageal oximetry is accurate in low-perfusion states. With an average Svo₂ of 74%, the 95% confidence interval for under- and overestimation is 6% and 7.8%, respectively. Because all measurements were made in hemodynamically stable patients, the errors may be increased in clinical practice. Second, although we detected no adverse events, an insufficient number of patients were studied to confirm its safety. Assuming a binomial distribution for the incidence of adverse events, the upper limit for the probability when no adverse events have been observed in 20 patients is 0.19 ⁽¹⁶⁾. Therefore, we can state with only 95% confidence that the technique is safe in at least 81% of patients. Until more data are available, we cannot recommend ventricular oximetry for general use. However, both TEE ⁽¹¹⁾ and esophageal oximetry ⁽⁷⁾ have an excellent safety record. Finally, our device was homemade, with the oximeter probe taped to the endoscope, but it would be possible to incorporate the oximeter into the wall of the endoscope. It should also be possible to miniaturize the apparatus, because high-resolution echocardiographic imaging may not be required to direct the probes toward the ventricles.

We conclude that TEE-guided LV and RV oximetry is feasible in hemodynamically stable anesthetized patients and provides similar readings to Sao₂ and Svo₂ samples. The technique merits further investigation.