Perioperative Support

Indications for ECMO support surrounding cardiac surgery include preoperative stabilization, failure to wean a patient from CPB, postoperative low cardiac output syndrome (LCOS), and postoperative cardiopulmonary arrest (CPA).

Preoperative stabilization may be required in infants with profound cyanosis, cardiogenic shock, or severe pulmonary hypertension (e.g., infants with d-transposition of the great arteries).

ECMO use for failure to wean from CPB typically entails direct cannulation of the right atrium and the aorta via the median sternotomy (central cannulation). Outcomes of patients who fail to wean from CPB have been shown to be worse than outcomes of those who are cannulated onto ECMO for LCOS or CPA occurring after a period of stability. Reported rates of mortality reported after postcardiotomy ECMO in infants and children range between 29% and 61%. Some of the risk factors for mortality in patients supported on ECMO postoperatively are peak serum lactate level within 24 hours of ECMO support initiation, renal failure requiring hemodialysis during the ECMO course, sepsis, increased blood transfusion requirements, and ECMO duration.

Residual cardiac defects need to be considered as potential cause for postoperative low cardiac output. Transesophageal echocardiography, direct cardiac chamber pressure, and oxygen saturation measurements can assist in diagnosing residual cardiac defects. An intraoperative review of preoperative data and imaging can also be helpful.

Determining contraindications for perioperative ECMO support in infants and children who undergo cardiac surgery is a complex process and varies widely among centers providing cardiac surgical care. Some historically absolute contraindications have evolved into relative contraindications in recent years, including prematurity, pre-ECMO neurologic injury, multisystem organ failure, and severe coagulopathy. Heparinization of the circuit may worsen hemorrhage (e.g., intracranial hemorrhage, uncontrolled visceral bleeding), and thus ECMO may be deleterious in these cases. Weight less than 2 kg can pose technical challenges to ECMO cannulation and support. When health care teams are considering ECMO support, it is important to understand that ECMO is only a support modality and does not offer any therapeutic benefit. Thus children who benefit from perioperative ECMO support have a reversible reason for their acute refractory heart failure.

The mean duration of cardiac postoperative ECMO course was 6 days in neonates and 7 days in pediatric patients reported to the ELSO registry between 2009 and 2016. Longer ECMO runs have been associated with higher mortality. Patients unable to separate from ECMO after 5 to 9 days possibly have a nonreversible cause for cardiac failure and should be considered candidates for cardiac transplantation. Specialized care teams of surgeons, cardiologists, intensivists, nurses, and other providers need to thoughtfully evaluate options, consider patient and family values, and assess regularly the risks and benefits of continued extracorporeal support if there is no improvement in the patient’s condition.

Support for Nonsurgical Conditions

Nonsurgical cardiac conditions supported with ECMO include myocarditis and cardiomyopathy, refractory pulmonary hypertension, intractable arrhythmias with hemodynamic compromise (including from toxic ingestion), septic shock, cardiac trauma, and posttransplantation rejection.

Post–Cardiac Arrest Support

Use of ECMO for support of patients with CPA refractory to conventional CPR (ECPR) and of patients with profound cardiogenic shock following return of spontaneous circulation after CPA has been increasingly reported in the literature. The number of ECPR cases reported to the ELSO registry increased by 35% for neonates and 67% for children between 2009 and 2015, with a mean ECMO duration of 5 days. The AHA 2015 guidelines state that CPR with ECMO (ECPR) may be considered for infants and children with cardiac diagnoses who suffer in-hospital cardiac arrest in settings that allow “expertise, resources, and systems to optimize the use of ECMO during and after resuscitation,” with insufficient evidence to suggest for or against the routine use of ECPR for infants and children with noncardiac diagnoses who suffer in-hospital cardiac arrest (weak recommendation, very low-quality evidence). In large national and international registries, survival rates to hospital discharge range from 40% to 43%. In an analysis of the AHA Get With the Guidelines—Resuscitation (GWTG-R) registry, children with in-hospital CPR of 10 minutes or longer who underwent ECPR had improved survival to hospital discharge and survival with favorable neurologic outcomes (defined as Pediatric Cerebral Performance Category [PCPC] 1 to 3 or unchanged from admission), compared with children with in-hospital CPR of 10 minutes or longer who underwent continued conventional CPR. Duration of CPR has been linked to mortality and neurologic injury after cardiac arrest in several studies, possibly due to decline in quality and effectiveness over time. As such, many centers have implemented rapid-deployment ECMO or ECPR programs, to limit CPR duration and minimize no-flow or low-flow states. With creation of portable ECMO consoles and circuits and improved ultrasonography to facilitate cannula insertion, expansion of these programs in recent years has led to deployment strategies not only in intensive care units and operating rooms, but also in emergency departments and out-of-hospital settings. Pediatric as well as adult ECPR studies suggest that the risk of mortality increases with longer CPR duration, as does the risk of poor neurologic outcomes. On the contrary, even prolonged CPR events longer than 1 hour can be associated with favorable neurologic outcomes. These issues illustrate that maintaining good quality of CPR while awaiting ECMO cannulation is essential to achieving good outcomes after ECPR.

Biventricular Circulation

Pediatric cardiac ECMO patients have varying degrees of univentricular or biventricular failure, pulmonary hypertension, or pulmonary parenchymal disease. Thus as many as 97% of these patients are supported with venoarterial ECMO, in which venous blood is drained from the right atrium and returned to the aorta, thus providing both cardiac and respiratory support. In contrast, in venovenous ECMO venous blood is drained from the venous circulation and returned to the right atrium. Thus in venovenous ECMO, normal cardiac function is essential. The total systemic blood flow (cardiac output) represents the sum of the extracorporeal flow and the native cardiac output. As opposed to CPB, in which all blood return to the heart is drained through cannulas in the superior and inferior vena cavae, in venoarterial ECMO, only ≈80% of venous return to the heart is drained through the circuit. As long as the heart can pump even small amounts of blood and the aortic valve continues to open, the pulse pressure (difference between the systolic and the diastolic blood pressures) is maintained at ≈10 mm Hg. Severe heart failure without ventricular contractility can be complicated by accumulation of blood from bronchial and thebesian venous flow in the left chambers and subsequent increased pressure in the left ventricle, left atrium, and pulmonary circulation. When this problem is not addressed by an atrial septostomy or placement of a drainage cannula in the left atrium or pulmonary artery, there is increased risk for left ventricle distention, pulmonary edema, and pulmonary hemorrhage. Furthermore, stagnation of blood in the cardiac chambers and pulmonary circulation can lead to clot formation despite anticoagulation.

During venoarterial ECMO, pump flow and systemic vascular resistance are adjusted to maintain age-appropriate cardiac output and mean arterial pressure. Any increase in left ventricular afterload potentially impacting myocardial recovery can be mitigated by reducing systemic vascular resistance or by decreasing preload while maintaining coronary blood flow during venoarterial ECMO support.

In the postoperative cardiac patient requiring ECMO support the goal is to maintain lung inflation and function so that decannulation is not impeded by respiratory failure at the time of cardiac recovery. This may require higher levels of positive end-expiratory pressure and peak inspiratory pressure or alternate modes of ventilation such as high-frequency ventilation or airway pressure release ventilation (see Chapter 23 ). The oxygenator fraction of inspired oxygen (FiO ₂ ) and sweep flow rates are adjusted to maintain normal oxygenation and normal PCO ₂ .

The Univentricular Circulation

ECMO has been used in single-ventricle circulations at all surgical stages. Generally the ECMO flow required for patients with single-ventricle physiology is higher than for patients with biventricular physiology, in an effort to maintain the systemic and the pulmonary parallel circulations. As such, patients with Blalock-Taussig (BT) shunt may require extracorporeal flows of 150 to 200 mL/kg/min. The BT shunt is typically maintained open. Standard-sized shunts (3.5 to 4 mm) usually provide enough limitation to pulmonary blood flow that increasing ECMO flow rate most often provides adequate systemic blood flow. Occasionally the shunt needs to be surgically constricted, but this could lead to shunt thrombosis, and diminished flow through the shunt risks pulmonary parenchymal ischemia and potential irreversible lung injury. The position of the right common carotid cannula needs to be checked serially to avoid migration of the tip into the shunt and potential loss of extracorporeal cardiac output. Successful ECMO support in superior cavopulmonary (Glenn) and Fontan circulations has also been reported; these patients often require multisite venous cannulation to provide adequate venous decompression and pump flow.

Rapid-Deployment Extracorporeal Membrane Oxygenation

Institutions that use ECMO as the primary mechanical support device for emergent circulatory support have developed rapid-deployment strategies to decrease the time required to initiate ECMO support. One or two circuits primed with crystalloid are maintained sterile for a set period of time (usually no longer than 1 month), ready for use in the case of ECMO cannulation during CPR or ECPR. Older children (usually above 20 kg) can be placed on ECMO emergently using the crystalloid-primed circuit, with subsequent typed and cross-matched blood transfusion to reestablish normal hemoglobin concentration if needed. Younger children may require a blood-primed circuit primed with type O, Rh-negative blood to reduce hemodilution. Availability of surgeons, in-house intensivists, and surgical team and equipment are requirements for successful and timely ECPR cannulation. High-quality CPR needs to be maintained for the entire duration of the cardiac arrest, with minimal interruptions for cannula placement, in an effort to avoid hypoxic injury. Team training through multidisciplinary simulations has been shown to decrease time from onset of cardiac arrest to time on ECMO in real patients in a pediatric cardiac intensive care unit.

Extracorporeal Membrane Oxygenation Teams: a Multidisciplinary Approach

ECMO is a complex, high-risk, and resource-intensive process with unpredictable resource use. Cardiac ECMO centers maintain well-trained multidisciplinary teams, with a core set of cardiac surgeons, intensivists, ECMO specialists, nurses, and perfusionists, as well as consultants from areas such as cardiology, pharmacy, nutrition, transfusion medicine, hematology, pulmonology, and nephrology. Multidisciplinary ECMO training using medical simulation is gaining ground in many cardiac ECMO centers after growing evidence for improved technical skills and team performance.

Neurologic Monitoring and Neuroprotection

Critically ill neonatal and pediatric cardiac ECMO patients are at high risk for neurologic injury surrounding their ECMO course. As many as 14% to 36% of ECMO patients suffer acute neurologic injury surrounding their ECMO course, including hypoxic-ischemic injury, thromboembolic stroke, and intracranial hemorrhage. Neurologic injury during ECMO is associated with an 89% increase in risk of mortality.

The mechanisms of neurologic injury in ECMO patients are not yet well understood. Profound cardiopulmonary failure and/or cardiac arrest lead to hypoxic-ischemic injury before ECMO, rendering the brain vulnerable. The profound inflammatory state associated with critical illness is then compounded by exposure to the foreign surfaces of the extracorporeal circuit, which triggers a profound global innate immune response by activating the contact and complement systems and provokes perturbations in proinflammatory and prothrombotic pathways implicated in the pathogenesis of neurologic injury during ECMO. The device-induced proinflammatory state and endothelial cell dysfunction increase thrombin generation and platelet activation. Within the patient the device contributes to coagulation factor consumption, platelet exhaustion, and reduced platelet aggregation potential. Added to that is the need for systemic anticoagulation to prevent circuit thrombus formation, creating the common paradox of thrombotic events in the device with simultaneous bleeding in the patient. Thus patients are at risk for both thromboembolic and hemorrhagic intracerebral complications.

A recent systematic review of the literature found that neuromonitoring methods during ECMO are limited and supported primarily by grade 3B-4 evidence from small, single-center studies (median n = 28, interquartile range [IQR], 18-49). Daily transfontanellar cranial ultrasonography for infants with open fontanels has entered routine clinical practice. Near-infrared spectroscopy–based cerebral oximetry and electroencephalography (EEG) are used in many centers. The American Clinical Neurophysiology Society recommends the use of critical care continuous EEG in children on ECMO support. A recent prospective quality improvement project conducted in a single quaternary care pediatric center showed that electrographic seizures occurred in 18% of consecutive neonates and children on ECMO support, were more common in patients with LCOS, and were associated with higher mortality and unfavorable neurologic outcome at hospital discharge. Other neuromonitoring methods, including transcranial Doppler ultrasonography and the serial monitoring or plasma brain injury biomarkers are being investigated and showing early promising results for detection of impending neurologic injury during ECMO and for prediction of unfavorable neurologic outcomes.

The best neurologic monitoring strategy is to minimize the use of sedatives and neuromuscular blocking agents. Limiting sedation when possible (e.g., in the absence of severe pulmonary hypertensive crises) has many other advantages, including spontaneous breathing and ability to interact with family and staff.

Respiratory Support

Respiratory management of neonatal and pediatric cardiac patients on ECMO is aimed at maintaining the lungs open and avoiding ventilator-induced lung injury, while taking into consideration cardiopulmonary and device-pulmonary interactions. In general the goal for neonatal and pediatric cardiac patients on ECMO is to avoid atelectasis, mobilize secretions, and, if present, support resolution of pulmonary edema due to capillary leak after large-volume transfusions, post CPB, or post cardiac arrest. Pressure control ventilation is used with moderately high positive end-expiratory pressure of approximately 10 cm H ₂ O, rate of 10 breaths/min, and pressures adjusted to achieve 6 to 8 mL/kg tidal volume, limiting peak inspiratory pressures to 18 to 20 cm H ₂ O. Limiting sedation and allowing for spontaneous respirations with pressure support or extubation on ECMO are becoming more common, especially in older children and adolescents, and may allow for improved cardiopulmonary interactions.

Nutrition

There are few published articles on optimal nutrition during ECMO. Generally a combination of enteral and parenteral nutrition is provided based on the patient’s nutritional requirements. The benefits of enteral nutrition have been well documented in critical illness at all ages, with recent data suggesting that early parenteral nutrition may be detrimental in critically ill children. Enteral nutrition has been shown to be feasible during venoarterial ECMO in neonates and children, potentially rendering benefits and cost savings. Some of the reported barriers for uninterrupted enteral nutrition during ECMO are ongoing vasopressor requirement, fasting for therapeutic or diagnostic procedures, and high gastric residual volumes. None of the neonatal, pediatric, or adult studies of nutrition during ECMO have reported complications related to early initiation of enteral nutrition in these patients. Antacids are commonly used for prophylaxis against acute gastritis.

Renal Supportive Therapy

Acute kidney injury is present at ECMO initiation in 60% to 74% of neonatal and pediatric patients and present by 48 hours of ECMO support in 86% to 93% of patients. Renal support therapy, most commonly in the form of slow continuous ultrafiltration and continuous venovenous hemofiltration, is used in a majority of neonatal and pediatric ECMO centers.

In neonates and children with critical cardiac disease, acute kidney injury may precede ECMO cannulation due to low cardiac output states, cardiac arrest, or exposure to nephrotoxic drugs. Acute kidney injury represents the most common indication for renal support therapy in cardiac ECMO centers and can worsen during the ECMO course if diuresis or hemofiltration are employed aggressively in an effort to augment fluid clearance and advance the patient toward decannulation. Although fluid overload and acute kidney injury are associated with increased duration of ECMO support and with increased mortality before hospital discharge, controversy remains as to the optimal rate of fluid removal and the optimal timing of initiation of renal support therapy during ECMO. If patients have adequate support on ECMO and can diurese adequately using their kidneys (with reasonable diuretic stimulation), then it may not be necessary to use forms of filtration. When there is impaired kidney function (usually heralded by decreased urine output despite adequate ECMO flow and often associated with rising levels of creatinine and blood urea nitrogen), then other forms of fluid management need to be considered. In some patients there is diffuse capillary leak resulting in third spacing, and in these patients, urine output may not be adequate despite reasonable ECMO flows. Often these patients have normalizing lactates (suggesting appropriate systemic oxygen delivery from the ECMO circuit) and a low central venous pressure. As long as it is possible to maintain flows in these patients, the capillary leak should recover, and as they begin to mobilize their third-spaced fluid, their urine output should recover and respond to diuretics. In some institutions these patients are managed with peritoneal drainage catheters (simply to drain the ascites and take any potential “pressure” off the renal vessels). When ascites reaches the point that an abdominal “compartment syndrome” is present (which can be measured by an elevated bladder pressure), then insertion of a peritoneal drain is a good idea. Patients who have true renal failure with low urine output, rising creatinine level, and high central venous pressures will often require some other form of filtration and are good candidates for continuous ultrafiltration while on ECMO. The decision making around the best management of low urine output requires multidisciplinary collaboration because often these patients present with a variety of reasons for low urine output that may require differing management strategies—one “solution” does not necessarily fit all circumstances.

Infection Risk and Surveillance

The prevalence of hospital-acquired infections during ECMO support is estimated at 10% to 12%, higher than in non-ECMO critically ill patients. Meticulous attention to hand hygiene and adherence to hospital-acquired infection prevention bundles is required to reduce risk of infection in ECMO patients. Although there are no data to support the notion that antibiotic prophylaxis beyond the pericannulation period (24 to 48 hours post cannulation) reduces the risk of infections acquired during ECMO, prolonged courses of antibiotics are commonly used to provide infection prophylaxis in many ECMO programs. Antibiotics commonly used for infection prophylaxis include cefazolin and vancomycin. Rarely antifungal agents are used as part of antibiotic prophylaxis. Use of aminoglycoside antibiotics for infection prophylaxis has been shown to be associated with increased incidence of sensorineural deafness in ECMO survivors and should be avoided when possible. Care of the skin to prevent pressure ulcers and preserving mobility of joints are important aspects of ECMO patient care.

Anticoagulation and Blood Product Management

Upon initiation of ECMO support, exposure of blood to the foreign surfaces of the circuit triggers an inflammatory response with activation of the coagulation pathway and blood elements (leukocytes, platelets) and thrombin activation, leading to a hypercoagulable state in the device. Anticoagulation to maintain patency of the circuit therefore needs to be balanced with the risk of bleeding within the patient. Despite accumulating experience with anticoagulation management and ever-improving biocompatibility of ECMO circuits, hemorrhage and thrombosis remain important complications during ECMO support and can be life threatening.

ECMO centers report large variability in anticoagulation monitoring and management. In an international survey conducted in 2012, 97% of respondents reported using activated clotting times (ACTs) to monitor UFH anticoagulation. Typically heparin infusion rates range from 10 to 40 U/kg/h and are adjusted for ACT targets of 180 to 220 seconds. Studies have shown, however, that ACT is an indirect and often inaccurate reflection of heparin effect and can be outside the target range for many reasons. Therefore currently most centers also monitor activated partial thromboplastin time (aPTT) (94%), anti-factor Xa (aFXa) (65%), thromboelastography (TEG) (43%), and antithrombin activity (82%), at varying time intervals. In addition, fibrinogen, D-dimers and platelet counts are measured serially to provide a more complete assessment of the patient’s coagulation status. aFXa is a reflection of actual heparin levels, and a target range is usually 0.3 to 0.7 to ensure adequate heparinization. Concomitant antithrombin III (AT III) levels can indicate the appropriate substrate for heparin to function, and when AT III levels are low, additional AT III (either as fresh frozen plasma or as a commercially available product [Thrombate III]) can be provided. Monitoring aFXa and AT III levels with adjustments as necessary has correlated with outstanding outcomes with respect to coagulation management in ECMO patients and is currently our preferred anticoagulation protocol.

ACT has several limitations; it is a global test of whole blood coagulation response, it has limited reproducibility, and results differ for different devices. ACT can be influenced by factors such as the presence of heparin, platelet count, platelet dysfunction, hyperfibrinogenemia or hypofibrinogenemia, other coagulation factor deficiencies, hypothermia, or hemodilution. In addition, ACT has poor correlation with aPTT and aFXa during neonatal and pediatric ECMO, and some have proposed that aFXa may be preferable as a marker for degree of anticoagulation during ECMO. In a single-center review comparing an anticoagulation protocol primarily driven by ACT versus aFXa and that included TEG and antithrombin measurements, Northrop et al. demonstrated significant reductions in blood product transfusions, cannula and surgical site bleeding, as well as improved circuit life with the aFXa-driven protocol.

Antithrombin measurement and administration have become common practices in neonatal and pediatric ECMO, despite the lack of robust evidence on efficacy and safety of both pooled and recombinant formulations. Antithrombin pharmacokinetics and pharmacodynamics during ECMO are poorly described. Data from single-center studies suggest that, in selected populations, antithrombin replacement can decrease exposure to blood products and improve efficiency of anticoagulation with unfractionated heparin (UFH) at least temporarily. There is still significant debate and controversy, however, regarding the safety of antithrombin administration, as well as contradictory reports of adverse events associated with antithrombin administration that will require further prospective study.

Alternatives to UFH, including direct thrombin inhibitors such as bivalirudin and argatroban, are being increasingly used in ECMO patients of all ages; the use of these agents in pediatrics patients has been reported in those who develop heparin resistance or heparin-induced thrombocytopenia. Some adult ECMO centers use bivalirudin as the first agent for anticoagulation, with significantly simplified anticoagulation management protocols.

Postoperative neonatal and pediatric cardiac patients on ECMO are at high risk for bleeding. Minor local bleeding such as bleeding at the cannulation site can be controlled with local measures, correction of coagulopathy via plasma, cryoprecipitate, or platelet transfusions and reducing the level of anticoagulation used. Massive bleeding requires prompt blood product administration, including platelets for a goal of 75,000 to 100,000/mm ³ , and administration of antifibrinolytics such as epsilon-aminocaproic acid. UFH infusion rate can be decreased or interrupted in the contemporary era of surface-modified ECMO circuits in cases of severe bleeding, although the risks and benefits of doing so must be discussed at the bedside, taking into careful consideration the cause of bleeding, thrombus burden in the circuit, and each institution’s circuit specification. The use of activated factor VII to control bleeding on ECMO has been reported from some centers. Access to a fresh primed ECMO circuit should be readily available when ECMO bleeding management requires stopping anticoagulation and administration of prothrombotic agents.

Even in the absence of active bleeding, red blood cells and platelets may suffer shear injury during ECMO, with subsequent hemolytic anemia and thrombocytopenia. Thus transfusion of red blood cells may be needed to maintain hemoglobin concentrations that optimize oxygen delivery (typically 10 g/dL), and platelet transfusions may be needed to maintain platelet counts deemed safe for avoiding spontaneous hemorrhage in the face of ongoing anticoagulation (typically 75,000 to 100,000/mm ³ ). Plasma or cryoprecipitate is administered for goal fibrinogen concentrations of 100 mg/dL, or 150 mg/dL before surgical procedures.