Advances in the field of assisted ventilation include computer-assisted ventilators with an ever-increasing number of modalities, bedside pulmonary function measurements, and graphics and drugs to modulate the pulmonary vascular bed and alveoli and manage the complications of ventilation. Despite the use of these advances, some neonates fail to improve and die of respiratory failure unless alternative treatment is available. Persistent pulmonary hypertension of the newborn (PPHN) and congenital diaphragmatic hernia (CDH) may not respond to assisted ventilation and require alternative therapy.
In addition to nonsurvival, many children who receive aggressive ventilatory support continue to experience unacceptably high rates of ventilator-induced pulmonary complications: barotrauma including pulmonary air leak, bronchopulmonary dysplasia, and chronic lung disease. These subsequent complications of ventilation impose increased rates of morbidity and mortality even after the newborn period. For all these reasons, judicious use of cardiopulmonary bypass techniques for temporary respiratory support of selected term and near-term newborns is needed and useful ( Figs. 40-1 and 40-2 ).
This technique is most frequently referred to as extracorporeal membrane oxygenation (ECMO), a name that unfortunately puts an undue emphasis on the role of oxygen support. Equally important may be the role of carbon dioxide extraction, cessation of toxic ventilatory settings, and cardiovascular support. Consequently, these processes of support may be referred to instead as extracorporeal life support (ECLS). ECLS expands the definition of extracorporeal support to include many newer cardiac support systems such as ventricular assist devices and pumpless extracorporeal support with membrane lungs only. In reality, both terms are frequently used when discussing neonatal support systems.
History of Cardiopulmonary Bypass
Artificial maintenance of circulation was pioneered by John and Mary Gibbon beginning in 1934 ; it was first reported in 1937 but not used in a widespread fashion by cardiac surgeons until the 1950s. It was soon discovered that if used for more than 1 to 2 hours, the device itself was lethal because of protein denaturation, which was thought to be caused by the gas exchange device. This finding led to the use of the biologic lung as the oxygenator for extracorporeal circulation, as described by Lillehei and colleagues. Because the major problem with artificial circulation was the oxygenator, many new devices were developed, including the membrane oxygenator and the bubble oxygenator, which became the standard for cardiac surgery.
During attempts to use these oxygenators for prolonged bypass, it was noted that the oxygenator, which directly exposes blood to oxygen (O 2 ), damages cells and proteins. Manifesting as coagulopathy and anemia, this effect is apparent within a few hours of initiating bypass. The large reservoir used for oxygenation also complicated management of volume and necessitated complete suppression of coagulation in the low-flow component.
Development of Membrane Oxygenators
Streamlined units that had no reservoir and incorporated a membrane oxygenator instead of a bubble oxygenator eliminated the direct blood–gas interface ( Fig. 40-3 ). The first membranes were made of polyethylene and Teflon but required large surface areas for adequate oxygenation. In 1957, Kammermeyer first reported the excellent gas transfer properties of a polymer of dimethylsiloxane, commonly known as silicone ( Fig. 40-4 ). This led to the development of many oxygenators and to the first trials in infants.
Once these membrane lungs and the circuits that bring blood to and take blood from them become coated with a protein monolayer, the blood is no longer in direct contact with a thrombogenic foreign surface. This allows for prolonged gas exchange with minimal cellular trauma, as well as elimination of the reservoir and the use of high-dose anticoagulation. For most patients, bleeding events are reduced and manageable.
Development of a Pump
In the development of neonatal ECMO, most devices were readily adapted from devices already in use by the cardiac surgery teams. Consequently, multiactivated Sigma motor pumps were used initially. Soon thereafter, roller pumps gained popularity because of their reliability and ease of use. With these devices, blood-conducting tubing is compressed, and the fluid is forced forward. To prevent increased hemolysis of red blood cells, partially occluding systems are used. More recently, the centrifugal pump has gained popularity. This type of pump has the advantages of low hemolysis, usability over a wide range of flows, and little risk for air pumping (air embolus); thus it has become the most common form of pump currently utilized in most ECMO centers.
The great disadvantage of all of the pumps in use as of this writing is the lack of pulsatile blood flow to the patient. This variation from the normal cardiac flow has physiologic effects on the end organs. Consequently, the use of venovenous bypass whenever possible with preservation of endogenous pulsatile flow may confer advantage to the sick neonate (see further discussion of venovenous cannulation later in this chapter).
The last major problem to be overcome in the quest to perform successful bypass in the neonate was vascular access. Early investigators used umbilical vessels, which did not provide adequate flow for substantial respiratory and cardiac support. Later, investigators cannulated the internal jugular vein and the common carotid artery. These sites allow sufficient flow to permit near-total cardiopulmonary bypass if needed. The successful solution of these multiple problems enabled Bartlett and his colleagues to complete the first successful application of ECMO for respiratory failure in a neonate in 1975.
Physiology of Extracorporeal Circulation
The membrane lungs in use as of this writing are the diffusion, hollow-fiber membrane oxygenators. Unlike other hollow-fiber oxygenators, the fibers are made of poly-4-methyl-1-pentene and are nonporous. These membranes are therefore a “true” membrane lung, as was the older version of the silicone membrane (no longer manufactured), and will not develop a plasma leak like the microporous hollow fiber oxygenators. Oxygen and carbon dioxide (CO 2 ) diffuse across the membrane at the molecular level ( Fig. 40-5 , shown for a silicone membrane; hollow-fiber membrane is essentially the same process). The gradient for O 2 diffusion across the membrane is the difference between the O 2 content in the ventilating gas and that in the venous blood of the patient.
Oxygen and Carbon Dioxide Transfer
The red blood cells closest to the membrane fibers become saturated with oxygen first, and the local partial pressure of O 2 (P o 2 ) increases. Dissolved O 2 then diffuses deeper into the blood film, saturating more red blood cells. The bundled fibers provide adequate surface area for gas exchange. For complete saturation of the blood film to occur, it must remain in contact with the membrane long enough for O 2 to diffuse to the center of the film. For any given membrane lung, the amount of venous blood that can be completely saturated is a function of the O 2 content of the venous blood returning to the membrane and the amount of time spent in the membrane. As the flow increases, blood spends less time in the membrane. Oxygen transfer increases in proportion to the flow rate until a limitation to O 2 transfer is imposed by the thickness of the blood film. When venous blood entering the membrane is 75% saturated, the flow rate at which blood leaving the membrane is 95% saturated is termed the rated flow of that device, a number that allows for standardization of various membrane lungs ( Fig. 40-6 ). If it is assumed that the membrane is large enough, the amount of O 2 that can be delivered is dependent on the blood flow available, not on the capacity of the membrane to transfer O 2 .
Carbon dioxide is much more diffusible through plasma than O 2 , and CO 2 transfer is limited by its diffusion rate across the membrane. The newer hollow-fiber membranes have such excellent gas transfer at low gas flow rates, 0.25 to 0.5 L/min, that it is unusual to require addition of CO 2 gas to the membrane gas mixture going across the membrane, as was usual with the older silicone membrane lung. Because CO 2 transfer is independent of blood flow but dependent on the surface area of the membrane, an increasing partial CO 2 pressure (P co 2 ) can be a sensitive indicator of loss of surface area and oxygenator function, which generally indicates clot formation or water in the gas phase.
Blood flow to the membrane is limited by the total circulating blood volume and the diameter of the venous catheter. The system must allow at least 120 mL/kg/min of flow to achieve support of cardiorespiratory function. The ECMO circuit is designed to permit this blood flow volume, with the membrane lung having a greater rated flow.
There are two critical questions in the application of ECMO: (1) Which patients can be helped, or alternatively, who is likely to die without this therapy? (2) When should this therapy be instituted? Because this is an invasive procedure, great effort to determine appropriate entry criteria for it should be undertaken before its institution.
The major criterion for ECMO selection is that the disease process must be reversible, usually within 2 to 4 weeks. ECLS beyond this time is difficult but has been successfully done for up to 2 months. Disease processes that lend themselves to ECMO therapy include meconium aspiration syndrome, pneumonia, neonatal sepsis, primary and secondary PPHN, CDH, perinatal asphyxia, respiratory distress syndrome, barotrauma with air-leak syndrome, and perioperative support of newborns with congenital cardiac lesions. ECMO has expanded from the newborn population to older children and adults ( Fig. 40-7 ).
In general, congenital cardiac defects can be identified and corrected without the need for ECMO. One cardiac condition that causes PPHN and may mimic some of the other conditions listed above is total anomalous pulmonary venous return. Unless there is an associated intracardiac defect, the heart itself may appear normal on a two-dimensional echocardiogram. The common pulmonary venous channel and absence of pulmonary veins entering the left atrium can only sometimes be demonstrated with noninvasive techniques. If the anomalous pulmonary venous drainage is obstructed, most commonly in the infradiaphragmatic variety, then pulmonary hypertension results.
ECMO may be useful for perioperative stabilization of very seriously ill infants with this condition and allows for completion of the workup and preparation for surgery in a more hemodynamically stable patient. ECMO can also be used as a ventricular assist device in the management of infants with perioperative ventricular failure, often allowing babies who would otherwise be unable to come off operative bypass to survive.
In theory, selection criteria for ECMO would allow providers to predict which patients will not tolerate traditional therapy before they develop life-threatening complications or irreversible lung injury. These patients would escape to ECMO at the point when the risks of conventional therapy outweighed the risks of ECMO. In practice, however, it has been difficult to determine this set point. Box 40-1 describes criteria used for scoring systems at a number of centers. While variation exists, these systems at the least provide guidance for when a discussion of ECMO use should occur.
A–aD o 2 greater than 610 × 8 hours or greater than 605 × 4 hours, if PIP is greater than 38 cm H 2 O
Oxygen index greater than 40
Acute deterioration with PaO 2 less than 40 × 2 hours and/or pH less than 7.15 × 2 hours
Unresponsive to treatment: PaO 2 less than 55 and pH less than 7.4 × 3 hours
Barotrauma (any four concurrently)
Pulmonary interstitial emphysema
Pneumothorax or pneumomediastinum
Persistent air leak for more than 24 hours
MAP greater than 15 cm H 2 O and subcutaneous emphysema
Postoperative cardiac dysfunction
Bridge to cardiac transplantation
Prolonged severe hypoxia
Prolonged mechanical ventilation for longer than 7 days
Structural cardiac disease
History or evidence of ischemic neurologic damage
Lack of parental consent
Lack of parental consent
Inadequate conventional therapy
Weight less than 2000 g
Gestational age less than 35 weeks
Contraindications to anticoagulation
Severe pulmonary hemorrhage
IVH grade II or greater
Prolonged mechanical ventilation longer than 7 to 14 days
History of severe asphyxia or severe global cerebral ischemia
Lethal genetic condition or unrelated fatal diagnosis (trisomy 13, trisomy 18, untreatable malignancy)
Untreatable nonpulmonary disease, significant untreatable congenital cardiac malformation or disease
ECMO, Extracorporeal membrane oxygenation; IVH, intraventricular hemorrhage; MAP, mean airway pressure; PIP, peak inspiratory pressure.
Alveolar–Arterial Oxygen Gradient
One of the original, and therefore the oldest, predictors of mortality in the neonate with respiratory failure is the alveolar–arterial O 2 gradient:
A-a DO 2 = P AO 2 − Pa O 2
The PaO 2 is measured directly from a postductal arterial blood sample, and the P ao 2 can be calculated from the alveolar air equation:
P AO 2 = PI O 2 − PA CO 2 / R + Pa CO 2 × Fi O 2 ( 1 − R ) / R