The underlying principle of ECMO relies on the premise that the patient has a reversible disease process that can be corrected with either treatment or rest, and that either will occur in a defined period of time. Prolonged exposure to high-pressure mechanical ventilation with high concentrations of oxygen can have a traumatic effect on the lungs, and in newborns frequently leads to the development of bronchopulmonary dysplasia (BPD). It has been suggested that BPD can result from high levels of ventilatory support for as little as 4 days or less. The pulmonary dysfunction that follows barotrauma and oxygen toxicity associated with mechanical ventilation typically requires weeks to months to resolve. Therefore, patients who have been ventilated for a long time and in whom lung injury has developed are not amenable to a short course of therapy with ECMO. Historically, most ECMO centers have not accepted patients who have had more than 10 to 14 days of mechanical ventilation, owing to the high probability of established, irreversible pulmonary dysfunction. However, as with many historical contraindications to ECLS, these restrictions are evolving. Indeed, though duration of pre-ECLS mechanical ventilation correlates with mortality in all populations, survival upward of 40% after >14 days of mechanical ventilation has been reported in neonates and children, likely justifying ECLS even in these patients.

Echocardiography should be performed on every patient being considered for ECMO to determine cardiac anatomy and function. Treatable conditions such as total anomalous pulmonary venous return and transposition of the great vessels, which may masquerade initially as pulmonary failure, can be surgically corrected but may require ECMO resuscitation initially. Infants with correctable cardiac disease should be considered on an individual basis. Indications for ECMO support in infants with cardiac pathology are based on clinical signs such as hypotension despite the administration of inotropes or volume resuscitation, oliguria (urine output <0.5 mL/kg/h), and decreased peripheral perfusion. Also, ECMO is an excellent bridge to cardiac and lung transplantation.

Because of the invasive nature of ECMO, and the potentially life-threatening complications, investigators have worked to develop an objective set of criteria to predict which infants will have an 80% mortality without ECMO. Three clinical measurement systems have been developed and tested to assist in identifying patients that will benefit from ECMO support:

• 1.

  Oxygenation Index (OI) = (MAP × FiO ₂ × 100)/PaO ₂

where MAP = mean airway pressure. This index has been evaluated and found that an OI > 40 in three to five postductal gases is predictive of a mortality risk ≥80%. ^{, ,} Currently, most centers begin considering application of ECMO with an OI of 25 to reduce the barotrauma associated with high-pressure mechanical ventilation.

• 2.

  Postductal Alveolar-Arterial Oxygen Gradient [(A-a)DO ₂ ]

An (A-a)DO ₂ of 610 Torr or greater despite 8 hours of maximal medical therapy predicted a mortality of 79%.

• 3.

  Ventilation Index = (Respiratory Rate × PaCO ₂ × Peak Inspiratory Pressure)/1000

A ventilation index >40 and OI > 40 was shown to be associated with a 77% mortality risk. Also, the combination of peak inspiratory pressure ≥40 cmH ₂ O and an (A-a)DO ₂ > 580 mmHg was associated with a mortality of 81%.

These clinical measurement systems are useful to quantitate the degree of cardiopulmonary derangement, and subsequently categorize patients into candidates for ECMO or continued maximal medical therapy. However, the decision to initiate ECMO is often a clinical decision based on clinical judgment and the patient’s individual response to maximal medical therapy. Patients are commonly started on ECMO when they have failed maximal medical support, significant barotrauma is imminent, and they are considered to have good potential for complete organ recovery.

The classic contraindications to ECMO are listed below. As ECMO treatment evolves and technology advances, many of these traditional contraindications to ECMO are being challenged:

• 1.

  Estimated gestational age less than 34 weeks : The higher incidence of intracranial bleeding in premature infants has historically precluded the use of ECMO in neonates less than 34 weeks estimated gestational age (EGA). ^, However, recent data indicate that ECMO can potentially be used in infants as low as 29 weeks EGA with acceptable rates of intracranial hemorrhage (ICH) and reasonable survival. Ideally, the development of nonthrombogenic coating of circuit components would obviate the need for systemic heparinization and decrease the risk of using ECMO in premature infants (see discussion below).

• 2.

  Birth weight less than 2 kg : Technical considerations and limitations of cannula size restrict ECMO candidates to infants weighing at least 2 kg. ^, The smallest single-lumen ECMO drainage cannula is 8 French, and flow through a tube is proportional to the fourth power of the radius. Small veins permit only small cannulas, resulting in flow that will be reduced by a power of four. Neonates who weigh less than 2 kg provide technical challenges in cannulation and in maintaining adequate blood flow through the small catheters. However, as with EGA, this weight cutoff has been challenged, as survival of up to 40% can be achieved at birth weights as low as 1.6 kg.

• 3.

  ICH greater than grade II : Patients with small intraventricular hemorrhages (grade I) or small intraparenchymal hemorrhage can be successfully treated on ECMO by maintaining a lower-than-optimal activated clotting time, in the range of 180–200 seconds. These patients should be closely observed for extension of the intracranial bleeding. Patients posing a particularly high risk for ICH are those with a previous ICH, a cerebral infarct, prematurity, coagulopathy, ischemic central nervous system injury, or sepsis. Consideration of these patients for ECMO should be individualized.

Neonates with ICH of higher grades are at increased risk of extension of their hemorrhage with systemic heparinization. This remains true today, but the development of technologies that obviate the need for heparinization may allow the use of ECMO in neonates with preexisting ICH in the future. ^{, ,} In addition, our experience has suggested that ECMO can be applied when expected mortality without ECMO is significantly higher in neonates with grade II ICH. If that is the case, ECMO can be considered, with lower levels of anticoagulation cautiously applied. Promising work with a nitric oxide-based, nonthrombogenic circuit may obviate the need for systemic anticoagulation.

• 4.

  Bleeding complications : Infants with ongoing, uncontrollable bleeding or an uncorrectable bleeding diathesis pose a relative contraindication to ECMO. ^, Any coagulopathy should be corrected before initiating ECMO because the need for continuous systemic heparinization adds an unacceptable risk of bleeding.

• 5.

  Mechanical ventilation for longer than 7 to 10 days : Classically, mechanical ventilation has been associated with a higher incidence of BPD and irreversible fibroproliferative lung disease. The duration of pre-ECMO mechanical ventilation is being challenged as data from the ELSO registry demonstrate survival of 50%–60% after pre-ECMO mechanical ventilation of up to 14 days.

• 6.

  Cardiac arrest that requires cardiopulmonary resuscitation (CPR) : Many centers now consider patients who suffer pre-ECMO cardiac arrest candidates for support. Survival rates up to 60% have been demonstrated in neonates who suffer cardiac arrest prior to or during cannulation. ^, Predictably, good outcomes are associated with effective CPR during the resuscitation.

• 7.

  Conditions incompatible with meaningful life after therapy—profound neurologic impairment, congenital anomalies, or other conditions : Every effort should be made to establish a clear diagnosis before the initiation of ECMO. Infants with anomalies incompatible with life do not benefit from ECMO. In the past, patients with trisomy 13 or 18 were not offered ECMO, but practice patterns are evolving, and there is a range of opinions from centers and providers. ECMO is not a resource that is intended to delay an inevitable death. Many lethal pulmonary conditions such as overwhelming pulmonary hypoplasia, congenital alveolar proteinosis, and alveolar capillary dysplasia may present as reversible conditions but are considered lethal. However, with improvement in medical and surgical care, conditions once thought to be nonsurvivable require constant reassessment.

Pulmonary vascular resistance (PVR) is the hallmark and driving force of the fetal circulation. Normal fetal circulation is characterized by PVR that exceeds systemic pressures, resulting in higher right-sided heart pressures and therefore preferential right-to-left blood flow through fetal shunts. The fetal umbilical vein carries oxygenated blood from the placenta to the inferior vena cava (IVC) via the ductus venosus. Because of the high PVR, most of the blood that reaches the right atrium (RA) from the IVC is directed to the left atrium through the foramen ovale. The superior vena cava delivers deoxygenated blood to the RA that is preferentially directed to the right ventricle and pulmonary artery. This blood then takes the path of least resistance and shunts from the main pulmonary artery directly to the descending aorta via the ductus arteriosus, bypassing the pulmonary vascular bed and the left side of the heart. Therefore, the lungs are largely bypassed during fetal circulation.

At birth, with the infant’s initial breath, the alveoli distend and begin to fill with air. This is paralleled by relaxation of the muscular arterioles of the pulmonary circulation and the expansion of the pulmonary vascular bed. These effects lead to a rapid drop in PVR to below systemic levels that, in conjunction with clamping of the umbilical cord, causes the left atrial pressure to become higher than the RA pressure. The result is closure of the foramen ovale, and all venous blood flows from the RA to the right ventricle and into the pulmonary artery. The ductus arteriosus also begins to close at this time. Therefore, all fetal right-to-left circulation ceases, completing separation of the pulmonary and systemic circulations. Anatomic closure of these structures takes several days to weeks. Thus, maintaining systemic pressure greater than the pulmonary circulation is vital to sustaining normal circulation.

Failure of the transition from fetal circulation to newborn circulation is described as PPHN (persistent pulmonary hypertension of the newborn) or persistent fetal circulation (PFC). Clinically, PPHN is characterized by hypoxemia out of proportion to pulmonary parenchymal or anatomic disease. In hypoxic fetuses and infants, the proliferation of smooth muscle in the arterioles may extend far beyond the terminal bronchioles, resulting in thickened and more reactive vessels. In response to hypoxia, these vessels undergo significant self-perpetuating vasoconstriction. Although sometimes idiopathic, PPHN can occur secondary to several disease processes such as MAS, CDH, polycythemia, and sepsis.

Treatment for PPHN is directed at decreasing right-to-left shunting and increasing pulmonary blood flow. Previously, most newborns were treated with hyperventilation, induction of alkalosis, neuromuscular blockade, and sedation. Unfortunately, these therapies did not reduce morbidity, mortality, or the need for ECMO. ECMO allows for the interruption of the vicious cycle of pulmonary vasoconstriction and hypoxia. By providing richly oxygenated blood, ECMO promotes relaxation of the pulmonary vascular bed, allowing the pulmonary blood pressure to return to subsystemic values without the iatrogenic complications associated with overly aggressive conventional therapy.

Data recommending permissive hypercapnia and spontaneous respirations as principles of treatment for these children have been reported. Inhaled nitric oxide (iNO) has been shown to decrease ECMO utilization and improve both oxygenation and neurodevelopmental outcomes. Sildenafil has also gained popularity, demonstrating a mortality benefit in these patients when compared to placebo alone, though it has never established an advantage over iNO in clinical trials. Hyperventilation and neuromuscular blockade are not part of the treatment strategy. These strategies have decreased morbidity, mortality, and the need for ECMO in several centers.

Neonates with CDH are of particular interest to pediatric surgeons. These patients are plagued with varying degrees of pulmonary hypertension and pulmonary hypoplasia. Often, pulmonary insufficiency ensues with a vicious cycle of hypoxia, hypercarbia, and acidosis. This process must be interrupted by medical management, which has vastly improved over the past two decades with the use of permissive hypercapnia/spontaneous respiration, pharmacologic therapy, and delayed elective repair, often collectively referred to as “gentle ventilation” or “gentilation.”

Various other strategies have been tried to manage critically ill newborns with CDH. High-frequency oscillation may have a major role in forestalling respiratory failure when used as an initial strategy rather than as a rescue. Surfactant plays no more than an anecdotal role. Nitric oxide is frequently used as a vasodilator in the treatment of pulmonary hypertension in these patients, though evidence backing this practice is lacking. ^, However, it may be efficacious in a subpopulation of infants with CDH and pulmonary hypertension with normal left ventricular systolic function. Other pulmonary vasodilators such as epoprostenol, sildenafil, and iloprost are starting to demonstrate some efficacy in babies with CDH. ^, The primary indicator for ECMO in the CDH infant occurs when tissue oxygen requirements are not being met, as evidenced by progressive metabolic acidosis, mixed venous oxygen desaturation, and multiple organ failure. The other major indicator is increasing iatrogenic pulmonary injury.

The goal is to maintain preductal oxygen saturations >85%. Spontaneous breathing is preserved by avoiding muscle relaxants. ^, Sedation is used as needed. Meticulous attention to maintaining a clear airway is obvious but critical. Gentle ventilation with spontaneous respiration is initiated with intermittent mandatory ventilation (IMV), 30–40 breaths per minute, equal I/E time, inspiratory gas flow of 5–7 L/min, peak inspiratory pressure (PIP) of 20–25 cmH ₂ O, and positive end-expiratory pressure (PEEP) of 5 cmH ₂ O. The Fi O ₂ is selected to maintain preductal Sa O ₂ greater than 85%. Although there is some variation, gentle ventilation is consistently defined with limitation in peak inspiratory pressures and permissive hypercapnia. If this method of ventilation is not effective, as demonstrated by severe paradoxical chest movement, severe retractions, tachypnea, inadequate or labile oxygenation (preductal O ₂ saturations <85%) or Pa CO ₂ greater than 65–75 mmHg, or pH less than 7.20–7.25 then a new mode of ventilation is needed.

If gentle ventilation at the aforementioned parameters is unable to maintain adequate oxygenation and pH, high-frequency jet ventilation (HFJV) or high-frequency oscillatory ventilation (HFOV) can be instituted. HFJV provides smaller volumes (1–3 mL/kg) more often at a much higher rate (240–660 breaths per minute) and expiration is passive. Oxygenation is proportional to mean airway pressure and ventilation is proportional to amplitude (peak inspiratory pressure vs. PEEP). Jet pulsations produce high-velocity laminar flow, which can bypass airway disruptions. HVOF differs in that it delivers smaller tidal volumes (1–2 mL/kg) at an even faster rate (8–15 Hz). The lung is inflated to a static volume, and then oscillated around the mean airway pressure.

Of all the indications for ECMO in neonates, CDH has the worst prognosis; though it varies across individual institutions, survival as reported by both ELSO and the CDH study group remains at 50%. ^, Therefore, patient selection for ECMO in neonates with CDH is of particular importance. There are several prenatal markers that can help risk stratify CDH severity and predict the need for ECMO postnatally. The lung-to-head ratio (LHR) is measured by prenatal ultrasonography (US). ^, It is defined as the product of the orthogonal diameters of the contralateral lung divided by the head circumference. Severe pulmonary hypoplasia is considered when the LHR is less than 1.0 with liver herniation. The LHR is operator-dependent and can only be obtained in a narrow gestational window. Therefore, o/e (observed/expected) LHR was developed, which is accurate at any gestational age. ^, Many centers are also relying on fetal MRI to measure total lung volume to predict mortality in fetuses with CDH. The total lung volume can be compared to the predicted lung volume based on gestational age. This o/e total fetal lung volume (TFLV) has been reported as a better predictor of mortality and the subsequent need for ECMO. Using either o/eLHR or o/eTFLV, CDH can be prenatally stratified into mild, moderate, and severe risk categories. A value of <25% in either modality qualifies as severe and predicts increased mortality and ECMO utilization as high as 67%.

Whether a baby should first demonstrate some evidence of adequate lung parenchyma remains controversial. Some physicians believe the best method to evaluate pulmonary hypoplasia and predict outcome is to evaluate the patient clinically. This is assessed by having a best Pa CO ₂ less than 50 mmHg and a preductal oxygen saturation greater than 90% for at least 1 hour at any time in the clinical course. With these criteria, successful ECMO should yield an overall survival rate of 75% or better. If patients with lethal anomalies, overwhelming pulmonary hypoplasia, or neurologic complications are not included, survival approaches 85%. ^{, ,} At the other extreme, Kays et al. have demonstrated 55% survival in infants who had a best pCO ₂ greater than 100 and pH less than 7.0 during the initial resuscitation. He therefore suggests offering ECMO to all patients regardless of physiologic parameters. Our group at the University of Michigan recently reported an intermediate approach. Infants with potentially lethal pulmonary hypoplasia were identified prenatally based on an LHR <0.8 with liver herniation and a fetal MRI with O/E TFLV of <25%. These infants were resuscitated, and if the baby could not demonstrate a pH > 7, pCO ₂ < 100, preductal S aO ₂ > 80%, and paO ₂ > 40 (least important, as it will likely be postductal) on ventilatory support utilizing a PIP <25 on CMV or MAP <20 on HFOV, with appropriate sedation and optimization of blood pressure over the first 2 hours of life, we did not proceed to ECMO, but instead offered comfort care. If the baby met these criteria at any point prior to 2 hours, we proceeded directly to ECMO. However, this practice was subsequently abandoned as, when applied to the CDH study group, these criteria did not predict survival. Regardless, it should be noted that the most severe CDH patients who are offered ECMO will likely have significant long-term morbidity if they survive.

The proper ECMO modality in infants with CDH is also debatable. Most centers use venoarterial (VA) ECMO in CDH patients. However, 10- and 15-year reviews of the ELSO database have concluded that mortality is no different between venovenous (VV) and VA-ECMO. ^, Renal complications and inotrope use were more common with VV, but neurologic complications were more common with VA. Cannula size is the main limitation to VV-ECMO use in infants with CDH. With the introduction of the 13F Crescent right atrial double-lumen cannula, VV ECMO is possible in infants. However, the choice between VA and VV should largely be based on the experience of the surgeon and institution with each modality.

Studies demonstrate that 1%–4% of pediatric intensive care unit (PICU) admissions suffer a cardiac arrest. Survival to discharge for a patient who has an arrest in the PICU ranges from 14% to 42%. The ELSO data demonstrate that approximately 73% of extracorporeal cardiopulmonary resuscitation (ECPR) has been used for patients with primary cardiac disease. Overall survival to discharge in this population has more recently been reported as high as 49%, whereas survival to discharge after ECPR for patients without congenital heart disease is 41%. The American Heart Association recommends ECPR for in-hospital cardiac arrest refractory to initial resuscitation, secondary to a process that is reversible or amenable to heart transplantation. Conventional cardiopulmonary resuscitation (CPR) must have failed, no more than several minutes should have elapsed, and ECMO must be readily available. Future research needs to analyze long-term neurologic status among survivors and identify which patients will benefit the most, with as little morbidity as possible.

Approximately 3% of patients that are treated with ECMO will require a second course. Survival after a second course of ECMO appears lower than after a single course in neonates, but in pediatric patients, the survival rates are comparable to the first course. Negative prognostic indicators for second-course ECMO patients include patients with renal impairment, higher number of first-course complications, age older than 3 years, or a prolonged second course.

VA cannulation can be performed with proper monitoring in the neonatal ICU or PICU under adequate sedation and intravenous anesthesia. The infant is positioned supine with the patient’s head at the foot of the bed. The head and neck are hyperextended over a shoulder roll and turned to the left. A transverse cervical incision is made over the anterior border of the sternocleidomastoid muscle (SCM), one fingerbreadth above the right clavicle. The platysma muscle is divided, and dissection is carried down either through or anterior to the SCM, which is retracted to expose the carotid sheath. The sheath is opened, and the IJV, common carotid artery, and vagus nerve are identified ( Fig. 5.1A ). The vein is exposed first and encircled with proximal and distal silk ligatures. Occasionally it is necessary to ligate the inferior thyroid vein. The common carotid artery lies medial and posterior, contains no branches, and is mobilized in a similar fashion. The vagus nerve should be identified and protected.

The cannulation procedure. (A) The carotid sheath is exposed, the sternocleidomastoid muscle is retracted laterally, and the common carotid artery and the internal jugular vein are dissected free. (B) The patient is anticoagulated after the vessels are dissected and ligated cephalad. The arterial cannula is passed into the junction of the innominate artery and the aorta. The venous catheter is passed into the right atrium. (C) A polymeric silicone (Silastic) bumper is used to facilitate ligation of the cannulas. The two ligatures on each vessel are then tied together.

The arterial cannula (usually 10 French for newborns) is measured so that the tip will lie in the ascending aorta. This is approximately one-third the distance between the sternal notch and the xiphoid. The venous cannula (usually 12 French for neonates) is measured so that its tip lies in the midright atrium, which is approximately a centimeter proximal to the nipple. A 2-0 silk tie is placed approximately 2.5 cm from the tip of the arterial cannula and approximately 6.0–6.5 cm from the tip of the venous cannula in a typical 3.0 kg infant. The patient is then systemically heparinized with 50–100 U/kg of heparin, which is allowed to circulate for 3 minutes. With VA ECMO, the venous cannula is usually inserted first. The proximal IJV is ligated cephalad to the site selected for the venotomy. Gentle cephalad traction on this ligature helps during insertion of the venous catheter. A venotomy is made close to the ligature. To aid in cannulation, two 5-0 Prolene sutures can be placed on the anterior lip of the venotomy for retraction when introducing the venous cannula. The venous catheter is inserted and advanced into the RA and secured with two 2-0 silk ligatures over a vessel loop placed under the ligatures on the anterior aspect of the vein to protect the vessel from injury during decannulation. The obturator is removed and the cannula back-bled and clamped to prevent air embolus.

In preparation for arterial cannulation, the carotid artery is ligated cephalad. Proximal control is obtained with an angled vascular clamp, and a transverse arteriotomy is made near the cephalad ligature ( Fig. 5.1B ). To help prevent intimal dissection, two 5-0 Prolene sutures are placed on the anterior lip of the arteriotomy and used for retraction when introducing the arterial cannula. The cannula is inserted to its premeasured position and secured in a fashion similar to the venous cannula. A vessel loop is placed under the ligatures on the anterior aspect of the carotid to protect the vessel from injury during decannulation ( Fig. 5.1C ). Again, the obturator is removed, and the cannula back-bled and clamped.

Any air bubbles are removed from the cannulas as they are connected to the ECMO circuit, and circuit flow is initiated. The cannulas are then secured to the skin above the incision ensuring there are good sutures in the postauricular area. An adhesive pad such as Duoderm can be placed under the cannula, with sutures through this and skin, to prevent skin breakdown. The incision is closed with monofilament suture; a topical hemostatic agent can be placed prior to closure if deemed necessary. The cannula positions are confirmed by both chest radiograph ( Fig. 5.2 ) and transthoracic echocardiogram. The tip of the venous catheter should be positioned in the mid-RA, and the arterial catheter in the ascending aorta at least 1–2 cm above the aortic valve.

On this chest radiograph in an infant undergoing venoarterial ECMO, the venous cannula can be seen with tip in the midright atrium, and the arterial cannula is positioned in the aortic arch.

For cutdown two-cannula VV, the procedure begins similarly as with VA ECMO. Venous cannulation is performed via the right internal jugular (RIJ) vein for reinfusion with the venous catheter tip in the mid-RA (∼6 cm in the neonate). Venous drainage in VV ECMO is via a femoral venous cannula to the IVC, which can be inserted either by cutdown or percutaneously with US guidance. Historically, the two-cannula VV ECMO was rarely used in neonates due to the small size of the femoral vein. However, motivated in part by a lack of available neonatal DLVV cannulas, Lillie and colleagues demonstrated the feasibility of a two-cannula VV technique in infants. Due to difficulty with drainage from the smaller femoral vein in neonates, the circuit is then reversed, with right atrial drainage and femoral vein reinfusion. More recently, the introduction of the Crescent RA DLVV cannula, an unicaval dual lumen cannula, has again made DLVV ECMO possible in infants.

DLVV ECMO can be approached with either unicaval or bicaval jugular cannulation. With a double-lumen Crescent RA venous catheter, the tip should be positioned in the superior half of the RA with oxygenated blood flow directed toward the tricuspid valve. Tip position in the distal atrium risks the potentially catastrophic complication of right atrial perforation. Special consideration is required when using bicaval cannulas such as the Avalon double-lumen bicaval cannula or Crescent bicaval cannula. These cannulas have both proximal and distal drainage side-holes that must be positioned in the superior and inferior vena cavas, respectively, while the reinfusion port, which is located between them, must lie in the RA and direct the oxygenated blood flow toward the tricuspid valve. While this allows for efficient circulation, it also makes accurate positioning of the cannula essential. Directing the cannula tip into the IVC can be challenging, and echocardiography and fluoroscopy should be used during placement to maximize safety and minimize episodes of malposition. Fig. 5.3A shows the Avalon cannula with the distances between the drainage and reinfusion ports with different cannula sizes, highlighting the importance and difficulty of proper placement, especially in smaller patients. Appropriate positioning on chest radiography is shown in Fig. 5.3B . Given these constraints, we do not utilize the Avalon cannulas in neonates. Recommended cannula types and sizes are shown in Fig. 5.4 .

The Avalon catheter has drainage ports that lie in the superior and inferior vena cavae, while the reinfusion port lies in the right atrium, directed toward the tricuspid valve (A). The distances between these ports can be very small in the smallest catheters, making proper positioning difficult. The chest radiograph shows proper placement of an Avalon catheter with the tip ( arrow ) lying in the IVC (B).

Recommended catheter types and sizes for VA and VV ECMO.

Courtesy Dr. Michael McMullan, Seattle Children’s Hospital.

A challenging situation may arise when one attempts to cannulate a newborn with a right-sided CDH. Anatomic distortion of the mediastinum can lead to cannulation of the azygos vein or to impaired preload, which will then fail to provide adequate ECMO support. This is usually detected by poor pump function and echocardiography. In these patients, attempted manipulation of a malpositioned cannula is often wrought with failure. Solutions include emergent CDH repair or central cannulation.

The pediatric population (ages 2–18 years of age) presents a difficult and controversial scenario regarding VA cannulation. Due to concern about carotid ligation, some centers will cannulate these patients via femoral access. One potential problem with this approach is the North-South syndrome, which necessitates conversion to VVA ECMO with an additional right IJV reinfusion cannula to oxygenate the upper body. In addition, the arterial cannula is large and can either partially or completely obstruct antegrade arterial flow to the leg, especially in younger patients. This can result in distal limb ischemia, which can lead to sensory or motor deficits, tissue loss, or even limb loss. One potential way to avoid this problem is to provide antegrade flow via a percutaneously placed distal perfusion catheter (DPC) in the superficial femoral artery ; alternatively, a posterior tibial catheter can be used to provide retrograde limb perfusion ( Fig. 5.5 ). On the other hand, some centers continue to perform arterial cannulation via the carotid artery. A recent study supporting carotid cannulation in the pediatric age group found that carotid ligation is associated with a 5.1% rate of stroke, which was only a 1.4% increase when compared to noncarotid ligation approaches. Furthermore, the data suggested that the rate of stroke may actually decrease with age when other factors are adjusted.

This infant has been cannulated for ECMO using the femoral artery and vein. To prevent possible distal limb ischemia, antegrade flow has been provided via a percutaneously placed distal perfusion catheter.

Operations, such as CDH repair, can be performed while the child remains on ECMO, but one must account for the challenges associated with operations on an anticoagulated patient. Hemorrhagic complications are a frequent morbidity associated with an operation on ECMO, and these complications increase mortality. Strategies to avoid these problems include the following: the platelet count should be greater than 100,000/mm ³ , the ACT should be reduced to 170–200 seconds or anti-Xa to 0.1–0.3 U/mL, AT-III replacement should be avoided, and the cautery should be used liberally, along with meticulous operative technique and hemostasis. The fibrinolysis inhibitor aminocaproic acid (100 mg/kg) is administered by bolus 1–6 hours prior to incision and then infused continuously (30 mg/kg/h) for 24 hours. ^, For larger operations such as thoracotomy, consideration should be given to use of the argon beam coagulator for large raw surfaces and application of topical hemostatic agents.

As the patient improves, less blood flow is required to pass through the ECMO circuit, and the flow can be weaned at a rate of 10–20 mL/h if the patient maintains good oxygenation and perfusion. The most important guide to VA ECMO weaning is the Sv O ₂ . For VV ECMO, it is the Sa O ₂ . Regardless of the cannulation format, successful weaning is marked by stable acid–base balance and good urine output. Other criteria for successful weaning include a body weight close to the patient’s dry weight, a clear chest film, and adequate lung volumes as measured by the ventilator.

When weaning, flows should be decreased to 30–50 mL/kg/min with a minimum flow of 100 mL/min to avoid circuit thrombosis. Newer oxygenators have higher limits of allowable flow, which may limit full weaning. Adjustable shunts placed across the oxygenator allow higher overall flow to the oxygenator with a lower flow being delivered to the patient. Also, flow probes placed on the arterial cannula can be used to accurately guide weaning. Twenty-four hours prior to a trial, the ventilator settings are increased, and lung recruitment strategies are employed. Common strategies include increasing PEEP to 8 cmH ₂ O, peak inspiratory pressure to 20–26, respiratory rate to 25–35, and targeting tidal volumes of 4–6 mL/kg depending on the underlying disease process. If there is concern for persistent cardiac dysfunction or pulmonary hypertension, a “low-flow echo” can be performed after turning the ECMO flow to 100 mL/min, though even such low ECMO flows will affect findings on echo. At the time of the weaning trial, if the child tolerates the reduced flow, all medications and fluids should be switched to the vascular access on the patient instead of the circuit. In VA-ECMO, the patient can be trialed off ECMO by flushing and clamping the cannulas, with the circuit bypassing the patient via the bridge. If it is possible that the child may need to be returned to ECMO support, then the cannulas should be briefly unclamped every 10 minutes to “flash” them and prevent thrombosis. In VV-ECMO, flow can continue, but sweep gas is stopped to trial the patient off support. In either case, the patient is then observed for 2 to 4 hours. If this is tolerated, decannulation can be accomplished. In select high-risk patients, it may be wise to leave the cannulas in place with a continuous low-dose heparin infusion for 24 hours prior to decannulation.

Decannulation is performed under sterile conditions with the patient in the Trendelenburg position. With the use of a short-acting muscle relaxant to prevent air aspiration into the vein, ventilator settings should be increased. The venous catheter is typically removed first, and the jugular vein is ligated. Repair of the carotid artery is controversial. Short-term results demonstrate acceptable patency rates and equivalent short-term neurodevelopmental outcomes when compared with children undergoing carotid artery ligation. ^, Another study of neonates who underwent arterial repair found a 72% incidence of an occluded or highly stenotic right common carotid artery at 2 years of age. Similar to other studies, there was no significant difference in neurologic development when compared to controls. With more regular use of postdecannulation brain MRI, there is emerging data to suggest that carotid reconstruction may be associated with increased incidence of intracranial hemorrhage than carotid ligation in neonatal respiratory ECMO, with no difference in ischemic lesions. The impact of these findings on neurodevelopmental outcomes remains unclear.

Neurologic complications, including ICH, cerebral infarct, and seizures, have developed in 26% of infants and children on ECMO over 40 years of clinical use. These complications carry significant mortality when encountered in ECMO patients. Frequent neurologic examinations should be performed, and the use of paralytic agents are avoided. The neurologic examination should include evaluation of alertness and interaction, spontaneous movements, eye exams, the presence of seizures, fullness of the fontanelles, tone, and reflexes. Electroencephalography (EEG) may also be helpful. Cranial US should be performed on all neonates before initiating ECMO to identify those patients in whom significant ICH already exists. A retrospective analysis revealed that birth weight and gestational age were the most significant correlating factors with ICH in neonates on ECMO, though in premature neonates the use of inotropes and the incidence of mechanical complications appear more significant risk factors for ICH than gestational age. Once the patient is placed on ECMO, US is repeated during the first 3 days when indicated by the clinical condition. If the examination reveals a new, moderate (grade II) hemorrhage or an expanding ICH, ECMO is usually discontinued.

If ICH is suspected or detected on cranial US and deemed to be small, it is reasonable to maintain a low ACT (180–200 seconds) or antiXa (0.1–0.3 U/mL) with a platelet count greater than 125,000–150,000/mm ³ . Serial head US should be performed to monitor the progression of the hemorrhage.

The ECMO registry reports an 11.6% incidence of cannulation site bleeding and a 13% incidence of other surgical site bleeding. Contact of blood with the foreign surface of the circuit activates the coagulation cascade. Platelet count and function are also affected. With anticoagulation, bleeding risks associated with operative procedures are increased considerably, and meticulous hemostasis, often aided by topical hemostatic agents, is imperative. If bleeding from the cervical incision is greater than 10 mL/h for 2 hours despite conservative treatment strategies, exploration may be needed. Topical hemostatic agents placed in the cervical incision are frequently helpful in this setting.

Bleeding into previous operative sites occurs frequently and must be handled aggressively. A decreasing hematocrit, an increasing heart rate, a decline in the blood pressure, or inadequate venous return are signs of ongoing hemorrhage. Treatment includes blood product resuscitation, including coagulation factors if necessary. ACT parameters should be decreased to 180–200 seconds or anti-Xa to 0.1–0.3 U/mL and the platelet count maintained above 125,000/mm ³ . Antifibrinolytics, such as aminocaproic acid, can also help prevent bleeding. The use of recombinant-activated factor VII (NovoSeven, Novo Nordisk, Inc., Princeton, NJ) has been described in the management of bleeding unresponsive to conventional methods. This is an off-label use, and thrombosis is a significant concern. Frequently, one must evacuate the hematoma and explore for the cause, which is often needed in the postcardiac surgery patient with an open chest and central cannulation. If bleeding is not quickly controlled, decannulation and cessation of anticoagulation may need to be strongly considered.

ECMO patients develop a coagulopathy secondary to consumption of clotting factors by the circuit, and if severe, may mandate a circuit change. Disseminated intravascular coagulation (DIC) represents a much less common cause of coagulopathy in ECMO patients, occurring in approximately 4% of cases. DIC is characterized by the consumption of plasma-clotting factors and platelets, resulting in deposition of fibrin thrombi in the microvasculature and hemorrhage. Sepsis, acidosis, hypoxia, and hypotension are the primary risk factors of DIC.

A dramatic decrease in pulmonary hypertension is frequently seen after the initiation of ECMO, usually within the first 48 hours. In the presence of a patent ductus arteriosus (PDA), this change in pressure results in a left-to-right shunt and contributes to decreased oxygenation efficiency, pulmonary edema, and poor peripheral perfusion. Fortunately, the PDA usually closes spontaneously with fluid restriction and diuresis. The use of indomethacin should be avoided because of its adverse effects on platelet function. Rarely is PDA ligation required or indicated while on ECMO.

Oliguria is common in ECMO patients and is often seen during the first 24–48 hours. The capillary leak that occurs with ECMO initiation can cause decreased renal perfusion, and the nonpulsatile blood flow that occurs with VA ECMO may contribute as well. Once the patient is adequately volume resuscitated and fluid shifts have stabilized, the use of furosemide (1–2 mg/kg) can improve urine output. If the creatinine continues to rise, then renal US is recommended. Continuous hemofiltration, which can be added in-line to the ECMO circuit, or CRRT can also assist in managing the fluid shifts, hyperkalemia, and azotemia. Hemofiltration removes plasma water and dissolved solutes while retaining proteins and the cellular components of the intravascular space.

The incidence of hypertension during ECMO varies from 28% to as high as 92%. According to the ELSO registry, 13% of ECMO patients require pharmacologic intervention. One group reported that detectable ICH occurred in 44% of their hypertensive patients and clinically significant ICH developed in 27%. The patient should be assessed initially for reversible causes of hypertension, such as pain, hypercarbia, and hypoxia. Embolic renal infarction is another cause of hypertension. Medical management includes the use of nicardipine, hydralazine, nitroglycerin, and captopril.

The ELSO registry data up to January 2017 describes an 8% culture-proven infection rate in ECMO neonates and pediatric patients. This is remarkably low, considering the large surface area of the circuit, the duration of bypass, and the frequency of access to the circuit. Risk factors for infection include the duration of ECMO support, the length of hospitalization, and procedures performed before the initiation of ECMO or during the run. Fungal infections carry a significantly higher hospital mortality rate, and sepsis confers higher morbidity and mortality in neonates. Access to the circuit should be minimized and meticulous sterile techniques are important.