Methods

Beginning in January, 2001, we decided to investigate the effectiveness of a protocol using routine application of postoperative ventricular assist for patients undergoing first stage palliation for hypoplastic left heart syndrome at Doernbecher Children's Hospital, part of the Oregon Health and Sciences University, Portland, Oregon. Institutional Review Board approval was requested prior to publication of this manuscript and report of the information. We palliated 13 consecutive infants for hypoplastic left heart syndrome, or used the Damus–Kaye–Stansel reconstruction of the aortic arch with augmentation for variations of hypoplasia of the left heart. All patients were assigned to a protocol of neurologic testing at preoperative and various postoperative intervals, beginning at 4 months from the Norwood procedure. Our technique involved using pulmonary homograft material to augment the aortic arch, atrial septectomy, and placement of an aortopulmonary shunt between the branchiocephalic artery and the right pulmonary artery. All patients were cooled on cardiopulmonary bypass to 18°C. Reconstruction of the arch was performed during a period of deep hypothermic circulatory arrest with intermittent perfusion.²⁷ All patients were rewarmed, separated from cardiopulmonary bypass, and treated with a period of modified ultrafiltration. Atrial and neoaortic cannulas were then attached to a roller pump in 11 patients, or a centrifugal ventricular assist device circuit in 2 patients, and flows were slowly increased to 150–200 ml/kg/min. Protamine sulfate was administered and hemostasis was achieved. Sternums were left open, and covered with Eshmark (Oregon Health and Sciences University, Portland, Oregon). The patients were then transported to the intensive care unit. Heparin was restarted in the intensive care unit after several hours and once hemostasis was complete. Patients were ventilated normally with inspired oxygen between 35 and 100% to keep arterial oxygen in the range of 40–50 Torr, and arterial saturations at 85 to 93%. Carbon dioxide was normalized in the range of 30 to 40. The aortopulmonary shunt was left open in all cases. Lactic acid was measured as an indicator of adequate systemic perfusion, and diuresis was initiated by infusion of furosemide once hemodynamics were stable and bleeding was under control. Patients were kept on the ventricular assist device until the levels of lactic acid returned to normal. In many cases, this was also associated with a normal heart rate, significant diuresis, and the physical findings exam of good perfusion. All infants received Milrinone^®, which was begun during rewarming on cardiopulmonary bypass, and continued during the period of ventricular assistance. Ventricular assistance was generally discontinued when the patients were felt to be hemodynamically recovered with adequate systemic perfusion, as revealed by normal levels of lactate, and good diuresis. Flows through the ventricular assist device were decreased, and inotropic support with Dopamine was initiated if necessary. Cannulas were removed and chests were closed in the intensive care unit. If chest closure resulted in significant elevation of the central venous pressure, it was delayed for 1 to 2 days after decannulation.

Results

All patients were stable on the ventricular assist device without any attempts to balance the systemic and pulmonary circulation. Levels of lactate had normalized to less than 2 mmol/l by 1.8 ± 1.1 days following surgery. The average duration of ventricular assist device support was 3.1 ± 1.0 days, with a range of 2 to 5 days. Of the 13 patients, 12 survived, giving an hospital survival of 92%. There was one case of mediastinitis in a patient who has now gone on to a successful Glenn anastomosis. In 2 patients, it proved necessary to explore the mediastinum because of significant postoperative bleeding. In 2 other patients, we noted persistent hypoxemia on the ventricular assist device. These patients were taken back to the operating room for elective revision of the shunt. In the first case, the neoaorta was found to be compressing the origin of the branchiocephalic artery. In the second case, the shunt was shortened. Both infants improved immediately, and were weaned from the ventricular assist device in the operating room. There were no other significant complications. Of the 12 survivors, 7 have undergone the Glenn procedure, and all 7 patients had normal cognitive and motor function and development at the time of their Glenn procedure as assessed by the Mullen and Vineland scales.

Discussion

Routine use of a ventricular assist device in the operating room following the first stage of Norwood palliation is a novel concept. The hospital survival of our patients has been excellent, albeit that ours is a small series. Notably, however, this represents the initial experience with the Norwood procedure at a Hospital in which there was no experience prior to 2001. Of particular note is that the postoperative management is simplified when cardiac output is ensured by mechanical circulatory assistance. Patients can be ventilated normally, and usually benefit from some supplemental oxygen. We rarely reduced the inspired oxygen below 40% the first postoperative night, and some patients benefited from inspired oxygen at levels of 100%. Acidosis, when present, was easily managed by increasing the flow through the ventricular assist device. It was common that the distal perfusion was excellent within 12 h as measured by physical examination, and revealed by strong pulses in the feet with warm legs, decreasing lactic acidosis, and increased urinary output. The intensive care team did not have to respond urgently to changes in systemic oxygen saturation, or to drops in perfusion or blood pressure. No patient experienced a cardiac arrest while being maintained on the ventricular assist device.

With the shunt open, it was expected that oxygenation would remain excellent. In 2 patients, systemic arterial oxygen saturations remained lower than expected, but these infants remained stable as support from the ventricular assist device maintained an adequate systemic delivery of oxygen to the tissues. Hypoxemia whilst receiving ventricular assistance indicates inadequate flow across the shunt, since systemic output should be adequate, and ventricular assistance enables these patients to remain stable until the shunt can be revised. When necessary, the ventricular assist device ensures the stability of the patient during transport to the catheterization laboratory for further investigation.

Mechanical circulatory assistance also maintains adequate cerebral delivery of oxygen to these patients who have obligatory hypoxemia. The hypoxemic patient has limited ability to extract oxygen, and when the hypoxemia is severe, as can be encountered following Norwood palliation, the brain is particularly vulnerable to injury.¹⁶ Although only 7 of our patients have come to the Glenn anastomosis thus far, all have normal Mullen and Vineland scores, measures of early leaving and adaptive behavior, respectively. These data, although preliminary, are encouraging in light of previously published reports related to neurologic outcome.^{11, 13, 14}

The combination of normal neurologic outcomes and excellent hospital survival, combined with ease of postoperative management, makes the use of routine ventricular assistance intriguing. It is certainly not meant to be presented as a panacea. One patient died of sepsis several days after being weaned from the device. When patients are weaned from the device, it was common, in our experience, that they benefited from 1 to 2 days of inotropic support to support their hemodynamics. In many cases, closure of the chest after removal from the device was associated with an abrupt increase in central venous pressure. In these patients, we found it helpful to wait 1 to 2 days prior to closing the sternum, especially if the patients required inotropes at the time of discontinuation from ventricular assistance. It is not possible for us to know whether or not the use of the ventricular assist device increased cost to these patients. We do know that survival and neurologic outcomes are excellent. Routine ventricular assistance device following Norwood palliation is a novel approach that addresses the increased demands for cardiac output following the operative procedure. Outstanding hospital survival in inexperienced programs, or programs with previously average outcomes with the Norwood approach, might be easily achievable by adopting a strategy that increases cardiac output, rather than one which emphasizes the responsiveness of the intensive care unit to balancing a limited cardiac output. Neurologic outcomes for patients managed with this strategy appear to be normal, although more data relating to this issue are being collected in a comparative manner.