Myocardial Performance Index with Sevoflurane–Pancuronium versus  Fentanyl–Midazolam–Pancuronium in Infants with a Functional Single Ventricle

After approval from the Institutional Review Board of Baylor College of Medicine and Affiliated Institutions, Houston, Texas, and parental informed consent, patients aged 4–12 months with a functional single ventricle presenting for bidirectional cavopulmonary anastomosis were enrolled. Patients were excluded if they were receiving intravenous inotropic medications or β-adrenergic blocking agents or were mechanically ventilated preoperatively.

Patients were randomized to one of two groups by selecting the first of a lot of prelabeled cards representing each group. Randomization was concealed until used. The two groups were fentanyl–midazolam and sevoflurane. All patients received premedication with 0.05–0.3 mg/kg intravenous midazolam to achieve a sedated but responsive state. After application of standard monitors and recording of heart rate, oscillometric blood pressure, and arterial oxygen saturation by pulse oximetry (Spo²), a baseline transthoracic echocardiogram was performed with the patient breathing room air. Anesthesia was then induced (Narkomed NAD 6000; Draeger Medical Inc., Telford, PA) with either an inhaled agent using calibrated vaporizers, with 10 l/min oxygen, or a fentanyl–midazolam infusion over 1 min. Muscle relaxation was facilitated with 0.15 mg/kg pancuronium, and the trachea was intubated. Fractional inspired oxygen concentration was then reduced and maintained at 0.21, and delivered tidal volumes were set at 10–12 ml/kg, with a respiratory rate set to maintain the end-tidal carbon dioxide at 30–40 mmHg. The lowest possible mean airway pressure was maintained, and peak inspiratory pressure remained less than 25 cm H²O; a positive end-expiratory pressure of 0 cm H²O and an inspiratory:expiratory ratio of 1:2 was used. For the sevoflurane group, an age-adjusted minimum alveolar concentration (MAC) of 1 was achieved: end-tidal sevoflurane concentration of 2.6% for patients aged 4–6 months and 2.5% for patients aged 6–12 months,16using the technique of overpressure with 4–5% inspired sevoflurane concentration17to achieve the target end-tidal concentration as rapidly as possible. This was designated dose 1 for the sevoflurane group. End-tidal anesthetic, carbon dioxide (ETco²), and inspired oxygen concentration were measured at a side port on the elbow connector attached to the patient’s endotracheal tube with an infrared sidestream device calibrated weekly with standard gas mixtures. For patients randomized to fentanyl–midazolam infusion and maintenance, the same anesthetic machine and fresh gas flow rates of 10 l/min were also used. Drug infusion rates were calculated based on published pediatric pharmacokinetic data to predict two different plasma concentrations: 4 and 6 ng/ml for fentanyl, and 100 and 200 ng/ml for midazolam. To achieve the first of these concentrations, 18 μg/kg fentanyl was administered over 1 min followed by an infusion of 4.3 μg · kg⁻¹· h⁻¹, and 0.29 mg/kg midazolam was administered over 1 min followed by an infusion of 0.14 mg · kg⁻¹· h⁻¹. This was designated dose 1 for the fentanyl–midazolam group. This predicted plasma concentration is associated with sedation, hypnosis, and analgesia for surgery.18–23The inhaled anesthetic or the fentanyl–midazolam infusion was maintained at a constant end-tidal concentration or infusion rate at the dose 1 level for 10 min, and a second echocardiogram was performed and vital signs were again recorded. The sevoflurane group then had inspired concentration increased rapidly using overpressure to 5–6% sevoflurane to achieve 1.5 MAC end-tidal concentration. This concentration was dose 2 for the sevoflurane group. The fentanyl–midazolam patients were given a second infusion over 1 min equal to 50% of the first, and the maintenance rate was increased by 50%. This level was dose 2 for the fentanyl–midazolam group. After a 10-min period of equilibration at constant end-tidal sevoflurane or fentanyl–midazolam maintenance infusion rate, a final echocardiogram was performed, and vital signs were repeated.

All patients received maintenance intravenous fluids until the time of induction and were allowed to ingest clear liquids until 2 h before induction. Only maintenance intravenous fluids were administered during the study period. No vasoactive drugs were given except for the anesthetic agents. Each patient had a right internal jugular vein catheter inserted (without Trendelenburg positioning) immediately after tracheal intubation to measure central venous pressure. A radial artery catheter was placed as soon as possible, and arterial blood gas was sampled during steady state ventilation and hemodynamics. Any hemodynamic response perceived to be due to these procedures was allowed to subside before echocardiographic assessment; all measurements were made during periods of steady state hemodynamics. Rectal temperature was maintained at 36°–37°C.

Two-dimensional and pulsed-wave Doppler transthoracic echocardiography was performed by an experienced pediatric echocardiographer using an Acuson Sequoia (Siemens Medical Solutions, Redwood City, CA) or GE Vingmed Vivid Five (General Electric Healthcare, Milwaukee, WI) ultrasonic imaging system. The echocardiographer was blinded to the type of anesthetic administered. Studies at baseline, dose 1, and dose 2 were each obtained over 3–5 min during a period of stable heart rate and blood pressure. Studies were recorded on videotape for later off-line analysis, which was performed by pediatric cardiology fellows specializing in echocardiography or an attending pediatric cardiologist specializing in echocardiography. All measurements were obtained over three consecutive cardiac cycles, with the average of the values recorded. MPI was calculated from pulsed-wave Doppler signals of the atrioventricular valve inflow and ventricular outflow for each ventricle with a simultaneous electrocardiogram, as described by Tei et al .7(fig. 1). DTI was obtained using pulsed-wave tissue Doppler optimized with decreased velocity settings and increased sweep speed, with the sample volume positioned at the cardiac base in either the left or the right ventricular wall free at the level of the atrioventricular valve annulus (fig. 3). The peak S (systolic)–wave as well as E (early diastolic)–wave velocities were measured in meters per second. The A (late diastolic)–wave velocities were also measured; however, most of the A waves were fused together with the E waves, and thus these A-wave velocity measurements were invalid. Right and left ventricular wall DTI values were measured and analyzed separately. Pulsed-wave Doppler inflow velocities of the atrioventricular valves were also measured. In addition, flow across the aortic or neoaortic (in the case of patients who had undergone a previous Norwood operation) valve was calculated using the velocity–time integral (VTI) method: flow = VTI × heart rate × valve area.24Valve area in square centimeters was calculated using the formula: area =πr², where r is the radius of the valve, assuming a circular shape.

The MPI measurements were the primary outcome variable. We calculated a sample size based on the following assumptions: A 25% change in MPI would be clinically significant. The SD of MPI measurements in single-ventricle infants who have undergone the Norwood operation is reported to be approximately 33%.11Using an α level of 0.05 and a power of 0.80, we calculated that it would be necessary to study 28 patients, and we chose to study 30 patients to account for the possibility of increased variation in MPI measurement.

Data are reported as mean ± SD. Statistical calculations and analyses were performed using Sigma Stat version 3.0 (SPSS, Chicago, IL). Analysis of variance for repeated measures was used to compare parameters at the three dose levels within the same group. To compare parameters between the fentanyl–midazolam and sevoflurane groups at each dose level, a two-way analysis of variance with repeated measures was used. The Holm-Sidak method was used for post hoc  pairwise comparisons of the mean responses to the different treatment groups. Chi-square analysis was used to compare the number of patients in each group with a left or right ventricle, and the t  test was used for other patient data. P < 0.05 was considered significant for all tests.

Thirty-one patients were enrolled; however, one patient was removed from the study because necessary transthoracic echocardiographic views to obtain MPI measurements were not obtainable. All other patients had sufficient quality of echocardiographic measurements to calculate MPI at all three periods of the study. Sixteen patients received sevoflurane, and 14 patients received fentanyl–midazolam. One patient in each group had inadequate views to obtain DTI velocities. All patients had adequate views to obtain atrioventricular valve inflow E-wave velocities.

Patient data are reported in table 1. Fourteen patients had hypoplastic left heart syndrome. Other diagnoses included four patients with double outlet right ventricle; five patients with double inlet left ventricle; five patients with tricuspid atresia; one patient with Ebstein anomaly; and one patient with mitral atresia, dextrotransposition of the great arteries, and single right ventricle. Fourteen patients had undergone the Norwood procedure, six had undergone previous pulmonary artery banding, seven had a modified Blalock-Taussig shunt, and three had undergone aortic arch reconstruction. Two patients had undergone no previous procedures. The groups were not stratified into right versus  left ventricles, and the fentanyl–midazolam group had significantly more infants with right ventricular morphology. The partial pressure of oxygen while breathing an inspired oxygen concentration of 0.21 during the study was higher for the sevoflurane group. All patients remained in normal sinus rhythm throughout the study period. All patients completed the study protocol without adverse effect. One patient with hypoplastic left heart syndrome died 7 days postoperatively after an uneventful operating room and early postoperative course. He experienced third-degree atrioventricular block and ventricular fibrillation presumed secondary to coronary ischemia and could not be resuscitated. All other patients survived to hospital discharge in good condition, without major complications.

The hemodynamic and oxygen saturation data are reported in table 2. Heart rate increased with sevoflurane, decreased with fentanyl–midazolam, and was slower with fentanyl–midazolam at dose 1 and dose 2 when compared with sevoflurane. Mean arterial pressure decreased in both groups; Spo²increased at dose 1 and dose 2 in the sevoflurane group and was higher than the fentanyl–midazolam group at both anesthetic levels but not at baseline.

Echocardiographic measurements and calculations are reported in table 3. Baseline MPI was not different between the sevoflurane and fentanyl–midazolam groups. MPI did not change in either the sevoflurane or fentanyl–midazolam groups for either anesthetic level. Baseline MPI for all infants with a single right ventricle was 0.50 ± 0.15 and was 0.35 ± 13 for single left ventricle infants (P < 0.05). However, when analyzed separately, both left and right ventricular MPI did not change in response to an anesthetic agent, i.e ., the presence of a single right ventricle did not result in worse myocardial function when exposed to either anesthetic regimen. Atrioventricular valve E-wave Doppler velocity did not change in the fentanyl–midazolam or sevoflurane groups at either anesthetic level. DTI S-wave and E-wave velocities did not change in the sevoflurane group; however, S-wave velocity in the right lateral ventricular wall and E-wave velocities in both left and right lateral ventricular walls were decreased with fentanyl–midazolam at both anesthetic levels when compared with baseline. Outflow from the aortic or neoaortic valve decreased with fentanyl–midazolam but not with sevoflurane.

This is the first report of echocardiographic measurements of myocardial function in response to anesthetics among infants with a functional single ventricle. A novel method was used to assess myocardial function that is not dependent on ventricular geometry, potentially making it an important tool for further studies in this patient population. We measured MPI with fentanyl–midazolam–pancuronium and sevoflurane–pancuronium in these patients and found no change in global myocardial systolic and diastolic function with either regimen. We found the baseline MPI to be higher (worse function) in single-ventricle patients than in children with normal two-ventricle hearts. This reflects decreased function of a volume-loaded single ventricle before cavopulmonary anastomosis in which blood flow through both systemic and pulmonary circuits is ejected from the ventricle in parallel. This results in an increased ventricular end-diastolic volume, which may cause decreased ventricular function according to the Frank-Starling mechanism.25In addition, 20 of the 30 patients in the study had a single right ventricle, resulting in both an increased volume and pressure load relative to the normal right ventricle ejecting only into the pulmonary circulation. The MPI has been broadly validated in adult6–8and pediatric26patients with two ventricles using invasive, noninvasive, and functional comparisons. However, validation techniques and studies in infants with a functional single ventricle are limited.10 

Based on our earlier studies of two-ventricle patients with congenital heart disease in which we found that patients exposed to sevoflurane had mild depression of left ventricular ejection fraction at 1.5 MAC end-tidal sevoflurane, we expected to find worsened MPI among functional single ventricle patients.27,28 

There are several potential reasons for the finding that sevoflurane did not worsen MPI. The muscarinic receptor blocking effect of pancuronium affects the sinoatrial node, sympathetic ganglia, and adrenergic neurons.29,30Pancuronium also inhibits reuptake of norepinephrine by adrenergic nerves.31These actions result in both positive chronotropic and inotropic effects.32This may explain the lack of any effect on MPI, especially in the sevoflurane group, where we expected to see at least a small decrease.

Another reason why we did not observe myocardial depression from sevoflurane may have been decreased anesthetic uptake. Patients with right-to-left intracardiac shunting, which included all of our patients, exhibit delayed wash-in of sevoflurane.33We used high fresh gas flows and the technique of overpressure to overcome this limitation; however, the arterial concentration of sevoflurane was undoubtedly lower than the target MAC level, and we did not measure plasma concentrations of sevoflurane. Although our patients all had cyanotic congenital heart disease, the preoperative calculated pulmonary-to-systemic blood flow ratio at cardiac catheterization was 1.6:1 for the sevoflurane group, typical of volume-loaded single-ventricle patients before cavopulmonary connection. This increased pulmonary blood flow would offset the problem of decreased uptake of sevoflurane, provided that the pulmonary-to-systemic blood flow ratio did not change from that measured at catheterization. We attempted to ensure this by using the same inspired oxygen concentration (0.21) and airway management (tracheal intubation and positive-pressure ventilation with positive end-expiratory pressure zero and mean airway pressure minimized). By contrast, the fentanyl–midazolam group had a pulmonary-to-systemic blood flow ratio of only 1.1:1, probably explained by the differences in ventricular morphology and the fact that fewer patients in the fentanyl–midazolam group had two ventricular outlets than in the sevoflurane group (4 of 14 in the fentanyl–midazolam group, 9 of 16 in the sevoflurane group). A single ventricular outlet leads to pulmonary blood flow totally dependent on a small systemic-to-pulmonary artery shunt and thus often less pulmonary blood flow than those with two outlets. The fentanyl–midazolam group should have achieved the target plasma concentrations and equilibration of the intravenous agents faster than patients with normal cardiac anatomy. This should be true because of their right-to-left intracardiac shunts compared with patients without such shunts, because a portion of the dose bypasses the lungs and passes directly into the arterial blood.34,35Also, because some of our patients had decreased pulmonary blood flow (pulmonary-to-systemic blood flow ratio less than 1:1), some of the pulmonary sequestration of fentanyl may have been avoided, increasing the plasma concentrations.36Again, we did not measure plasma concentrations of fentanyl and midazolam.

By chance, 7 of the 16 patients in the sevoflurane group had a single right ventricle, versus  13 of the 14 patients in the fentanyl–midazolam group. The fentanyl–midazolam group did show a higher MPI at baseline; however, we could not demonstrate any worsening of MPI when all right ventricle patients were analyzed together. Previous studies have demonstrated that before cavopulmonary connection, function of single right versus  left ventricles in infants younger than 1 yr is similar. This is true whether assessed using MPI10or other echocardiographic measures of function, such as velocity of circumferential fiber shortening, wall stress, or change in end-systolic and end-diastolic areas.37Other studies of functional changes throughout staged Fontan reconstruction for single-ventricle lesions have combined patients with single right and left ventricles for data analysis because of the similarities of changes in geometry and function through the different operative stages.38Therefore, although it would have been ideal to stratify the randomization into right versus  left ventricle morphology, it is likely that the results of our study would not change.

E-wave atrioventricular valve inflow velocity also did not change with either fentanyl–midazolam or sevoflurane. DTI S-wave and DTI E-wave velocities did not change with sevoflurane, either. However, fentanyl–midazolam resulted in significant decreases in both DTI S-wave and DTI E-wave velocities at both anesthetic concentrations, which could be interpreted to mean impaired systolic and early diastolic longitudinal ventricular function with fentanyl–midazolam. Preservation of DTI S-wave and DTI E-wave velocities with sevoflurane may be due to decreased afterload with this agent, combined with preserved longitudinal shortening. In our previous studies,27,28sevoflurane decreased systemic vascular resistance (i.e ., afterload) more than fentanyl–midazolam. We could not calculate systemic vascular resistance in this study because all patients had the entire systemic output and at least the major portion of the pulmonary output ejected through the aorta. Again, the vagolytic effect of pancuronium combined with sevoflurane may explain the preservation of the DTI values with sevoflurane. Also, we could not measure late diastolic function (the A wave) by DTI because the A waves were fused with the E waves in the majority of patients as a result of high heart rates. Differential effects of the two regimens during this late diastolic phase of the cardiac cycle could be another explanation why MPI did not change with either sevoflurane or fentanyl–midazolam but the DTI values were different with fentanyl–midazolam. Because MPI is a measure of global systolic and diastolic function of the entire ventricle rather than an isolated myocardial wall segment,15it may be a more sensitive indicator of changes in global myocardial function in response to anesthetic regimens than DTI. In normal children, DTI velocities correlate best with age and ventricular end-diastolic volumes but have a variable and sometimes weak correlation with echocardiographic parameters such as shortening fraction and MPI.15As measured in this study, DTI is an isolated longitudinal measure of ventricular function, which does not account for circumferential fiber shortening, and thus gives different information than the MPI. DTI is also a new modality in pediatric patients, with most published reports and comparisons to other echocardiographic and clinical variables only in healthy children.14,39–44Therefore, the DTI technique awaits further validation in more children with abnormal hearts and with changes in function as measured by other parameters.

In our previous studies of two-ventricle patients, we measured systemic blood flow (i.e ., cardiac output), using the velocity–time integral method, which assumes a circular aortic valve annulus.27,28This assumption may be reasonable in patients whose aorta has not been surgically altered. In the current study, we measured flow across the aorta or neoaorta; however, 14 of the 30 patients had previously undergone the Norwood operation, and the reconstructed neoaorta may not have a circular annulus or the patient may have altered flow patterns (turbulence or regurgitation), introducing potential error into the echocardiographic calculation of flow. Aortic flow decreased with fentanyl–midazolam; most likely the slower heart rate with fentanyl–midazolam combined with the lower velocity–time integral (a surrogate for stroke volume) contributed to this change. Aortic flow was preserved near baseline levels with sevoflurane; presumably because the heart rate increased significantly and this compensated for the observed decrease in velocity–time integral.

In conclusion, we have used a new measurement of ventricular function, the MPI, to demonstrate that two commonly used anesthetic regimens, sevoflurane and fentanyl–midazolam, along with pancuronium, in doses commonly used for cardiac and noncardiac surgery, have no significant effect on myocardial function in infants with a functional single ventricle. Further study is required, particularly with sevoflurane, to measure arterial concentrations of the anesthetics to attempt to correlate them with myocardial function as measured by the MPI. Studies without vagolytic neuromuscular blocking agents would also be desirable. This would further elucidate whether these anesthetic agents, when given time to equilibrate, indeed have little or no effect on cardiac function in this important and difficult-to-treat patient population.

The authors thank Debora L. East, R.N. (Research Nurse, Division of Pediatric Cardiovascular Anesthesiology, Texas Children’s Hospital and Baylor College of Medicine, Houston, Texas), for assistance with patient recruitment and data collection.