Extracorporeal Membrane Oxygenation for Severe Pediatric Respiratory Failure

Pediatric ARDS

The literature has previously referred to pediatric ARDS as acute hypoxemic respiratory failure, acute lung injury, and simply ARDS. The 2015 Pediatric Acute Lung Injury Consensus Conference summary²² defined pediatric ARDS as a condition characterized by non-cardiogenic, new-onset pulmonary infiltrates in the setting of oxygenation deficit in patients ranging from infants through young adults. The guidelines specifically distinguished pediatric ARDS from the unique conditions seen in perinatal related lung disease and the pathophysiology related to persistent fetal circulation and rapidly changing pulmonary vascular resistance.²² No specific upper age limit was set, further emphasizing the similarities between a physically mature teenager and a young adult suffering from ARDS.

The specific criteria outlined for pediatric ARDS were derived from the American-European Consensus Conference adult definitions presented in 1994²³ and the subsequent large body of literature applying these definitions to the pediatric population. The updated Berlin criteria addressed important deficits related to previously missing consideration of applied mean airway pressure but also did not incorporate pediatric considerations.²⁴ The most significant addition to the Berlin criteria by the pediatric ARDS definition are the inclusion of noninvasive oxygen saturation measurements to characterize the degree of oxygenation deficit in those pediatric patients lacking arterial blood sampling. Inclusion of S_pO2/F_IO2²⁵ and oxygen saturation index ([mean airway pressure × F_IO2 × 100]/S_pO2)^26,27 along with P_aO2/F_IO2 and oxygenation index (OI = [mean airway pressure × F_IO2 × 100]/P_aO2) was based on studies demonstrating concordance between P_aO2/F_IO2 and S_pO2/F_IO2 ratios as well as OI and oxygen saturation index when S_pO2 was < 97%.

Using these oxygenation variables, degrees of pediatric ARDS severity were defined as mild (4 ≤ OI < 8 or 5 ≤ oxygen saturation index < 7.5), moderate (8 ≤ OI < 16 or 7.5 ≤ oxygen saturation index < 12.3), and severe (OI ≥ 16 or oxygen saturation index ≥ 12.3) in patients receiving invasive mechanical ventilation. For patients receiving noninvasive ventilation where mean airway pressure measurements are less reliable, oxygenation deficit criteria were defined using a P_aO2/F_IO2 ratio ≤ 300 or an S_pO2/F_IO2 ratio ≤ 264. This modification implicitly acknowledged the drawbacks of relying on the variations in bedside clinical decisions about the need for invasive arterial access or potential technical challenges in establishing invasive arterial access to define a clinical syndrome.

Defined using these specific pediatric considerations, pediatric ARDS represents as much as 4% of all pediatric ICU admissions²⁸ and at a population level occurs at an incidence of 2.0–12.8 cases/100,000 person-years with an overall mortality of 18–35%.²² Several risk factors for mortality have been described and can be categorized as either patient-specific or disease-specific (Table 1). Of the patient-specific risk factors, acquired or congenital immunodeficiency has demonstrated the highest mortality risk, in excess of 70%²⁹ and as high as 91% in patients with acute respiratory failure associated with bone marrow transplantation.³⁰ With respect to disease-specific risks, development of additional organ dysfunction syndrome and, in particular, multiple organ dysfunction syndrome have been demonstrated to be the most important mortality predictors.²⁹ Erickson et al²⁸ demonstrated that the odds ratios of death were highest with renal failure (odds ratio [95% CI] = 11.11 [3.95–31.26]), need for renal replacement therapy (odds ratio [95% CI] = 17.18 [3.64–31.26]), and multiple-organ failure (odds ratio [95% CI] = 8.34 [3.42–20.34]).

In addition to the presence of immunodeficiency or organ dysfunction, measures of ventilation and oxygenation also correlate with mortality risk. A ventilation index ([P_aCO2 × peak airway pressure × breathing frequency]/1,000) of > 65 by day 3 of illness had a positive predictive value for mortality of 90% in a small single-center retrospective study of 39 children.³¹ In another single-center study of 95 subjects with pediatric ARDS, alveolar dead-space fraction ([P_aCO2 − P_ETCO2]/P_aCO2) also correlated with mortality. Each alveolar dead-space fraction increase by 0.1 increased the mortality odds ratio by 1.92.³² Subsequent secondary data analysis demonstrated an alveolar dead-space fraction cut point of 0.23, with a 46% versus 13% mortality for subjects with alveolar dead-space fraction greater or less than this level, respectively.²² The OI has demonstrated the best discrimination of mortality risk with an area under the receiver operator curve of 0.747. An OI >16, representative of severe pediatric ARDS, appears to be a cut point, with mortality increasing from <23% for moderate pediatric ARDS to >30% for severe pediatric ARDS. Additionally, subjects with an OI >20 exhibited a 40% mortality.²² The same subject data set used to derive these OI cut points further demonstrated that each additional OI increase of 1 conferred an increased mortality odds ratio of 1.04.³³ Using a multivariate analysis, the combination of alveolar dead-space fraction and either OI or oxygen saturation index provided the best mortality prediction with an area under the receiver operator curve = 0.78, although this was not significantly better than using OI alone.³² Combining these 2 measures, an OI >20 and an alveolar dead-space fraction > 0.23 represent reasonable threshold values that identify patients whose mortality risk without ECMO may exceed the reported overall mortality associated with severe pediatric ARDS.

However, the decision to proceed with ECMO support for pediatric ARDS cannot be reduced to identification of preexisting immunodeficiency or developing organ dysfunction or to calculation of oxygenation and ventilation variables. These same variables that carry higher risk of mortality in pediatric ARDS without ECMO also have been associated with worse or unclear outcomes in patients undergoing ECMO support. A review of the ELSO registry through 2004 of patients requiring ECMO for respiratory failure who also had immune-compromised conditions had a somewhat lower overall survival to hospital discharge of 37% compared with the non-immunocompromised cohort with a survival of 58%.³⁴ Within this group of conditions, none of the 17 patients in the ELSO registry who had undergone bone marrow transplant and required ECMO for respiratory failure before 2004 survived. From 2004 to 2012, the ELSO registry reports 12 bone marrow transplant patients requiring ECMO, 3 of whom survived to hospital discharge (25%), perhaps reflecting improved overall critical care strategies as well as improved ECMO management.³⁵

The impact of organ dysfunction, especially acute kidney injury, before ECMO therapy or that develops while receiving ECMO therapy appears mixed. A single-center analysis of the impact of ECMO therapy on renal function in children demonstrated an improvement in pre-ECMO acute kidney injury soon after the initiation of extracorporeal support.³⁶ However, in addition to the renal toxic effects of the underlying disease process and renal hypoxia, acute kidney injury may also be caused by ECMO therapy itself. ECMO-related variables contributing to acute kidney injury include hemolysis and release of free hemoglobin, changes in hemodynamic flow characteristics while receiving ECMO, and impaired hormonal regulation.^37,38 A review of ELSO registry data from 1998 to 2008 for pediatric respiratory ECMO patients demonstrated an odds ratio for mortality of 1.7 with acute kidney injury that increased further to 2.5 if renal replacement therapy was administered.³⁹ In contrast, a single-center retrospective study demonstrated no difference in ECMO survival with or without the need for renal replacement therapy during the ECMO course.⁴⁰ However, the majority (67.4%) of these ECMO subjects had renal replacement therapy initiated due to fluid overload rather than renal failure (20.9%) and may misrepresent the relationship between acute kidney injury and ECMO survivability. Furthermore, the study recruited too few subjects to detect a mortality difference, requiring approximately 10-fold more subjects in order to be adequately powered.⁴¹ Whether acute kidney injury resulting from hypoxemic and inflammatory stress seen in pediatric ARDS can be prevented or ameliorated with improved oxygenation and ventilation provided by ECMO or whether ECMO contributes to development or worsening of acute kidney injury remains unclear.

The length of mechanical ventilation before the initiation of ECMO appears to also correlate with survival. Historically, the first ECMO trial in 1978 listed “duration of the precipitating pulmonary insult last[ing]…more than 21 days” as an exclusion criterion.² Subsequently, in 1993, Moler et al^42,43 determined a cut point of approximately 7 d of mechanical ventilation before ECMO initiation as an independently associated variable with mortality. More recently, a 2012 review of the ELSO registry demonstrated that decreased survival odds occurred when pre-ECMO length of mechanical ventilation exceeded 14 d.⁴⁴

Similarly, previously considered absolute contraindications for ECMO, such as recent trauma,⁴⁵ septic shock,⁴⁶ bone marrow transplant,³⁵ and diffuse alveolar hemorrhage,^47,48 each have been successfully treated with ECMO. As experience builds and techniques improve, the only absolute contraindications to ECMO are quickly becoming (1) known irreversible, terminal underlying disease with no viable cure and (2) uncontrollable hemorrhage despite maximal medical and surgical therapy. Ultimately, evaluating mortality risk of pediatric ARDS with or without ECMO is inexact at best and must consider not just the variables discussed above, but also the intensity and rate of clinical deterioration or lack of clinical improvement.

ECMO complications fall into 4 general categories related to (1) mechanical complications related to equipment failure and hemolysis from physical damage to the red cell; (2) bleeding and thrombolytic complications impacting the circuit, patient, or both; (3) infection; and (4) organ dysfunction (Table 2). The risk of these complications increases with longer ECMO duration although no data exist quantifying the degree of cumulative risk over time.

These ECMO-specific complications and other consequences of both ECMO therapy and the underlying disease process for post-survival morbidity must also be considered in the decision-making process. The earliest reports of long-term neurodevelopmental outcomes following ECMO support focused on survivors of ECMO applied in the neonatal population and compared the incidence of intellectual disability, learning disability, and cerebral palsy with healthy normal controls who had not required neonatal intensive care unit or special care nursery admission following delivery. A single-center cohort of neonatal ECMO survivors undergoing comprehensive neurocognitive assessment and parental and school survey assessments demonstrated major disability (mental or motor disability, sensorineural impairment, or seizure disorder) in 17 of 103 (17%) study subjects.⁴⁹ Major disability correlated highly with hemorrhagic or non-hemorrhagic intracranial lesions, although the relationship between these intracranial lesions and the ECMO course was not described. Parents of these ECMO-treated subjects also reported significantly more behavioral adjustment problems in 42% versus 16% of controls. Half of the ECMO subjects were classified as being at risk for school failure, over twice the rate of controls. This report, however, compared ECMO-treated subjects with healthy controls rather than critically ill neonates who did or did not receive ECMO. This weakness was addressed by the 7-y follow up of a randomized control trial of ECMO for neonatal respiratory failure.⁹ This analysis determined that cognitive ability, neuromotor performance, and health services utilization were similar in the ECMO and conventional treatment groups, suggesting that long-term neurocognitive and health issues were related more to the underlying disease process than to ECMO itself.

Few pediatric studies assessing the long-term impact on post-survival morbidity have been completed. Those that have been done have had limited subject numbers or have had indirect, less robust assessment methods. These pediatric reports also only provide overall incidence rates of disability without comparison with critically ill children not treated with ECMO. These studies have reported (1) a 16% prevalence of seizures and/or developmental delay and a 62% hospital readmission rate within a median of 1.2 y using a statewide administrative database and ICD-9 codes⁵⁰; (2) 38% mild or moderate disability based on the pediatric overall performance category scale at a median of > 5 y after ECMO based on parental telephone interview^51,52; and (3) 70% incidence of abnormal quality of life based on retrospective review of post-ECMO, pre-discharge medical records and prospective parental telephone survey.⁵³ Beyond a single report of a single pediatric patient rescued with ECMO for necrotizing pneumonia from pneumonia,⁵⁴ there are no published data describing the subsequent pulmonary morbidity and decrease in lung function, measured either formally via pulmonary function testing or informally with subjective assessment of exercise tolerance.

Consequently, the decision to proceed with ECMO in the setting of pediatric ARDS in a previously healthy child with no additional non-pulmonary organ dysfunction can be made by following serial measurements of a combination of OI or oxygen saturation index and alveolar dead-space fraction to determine clinical trajectory. However, in the face of concomitant immune deficiency or organ dysfunction, this decision is much more challenging with limited definitive evidence to guide the clinician and family. The question of long-term survival and disability also has limited data upon which to base life-and-death decisions. In this setting, realistic expectations among the medical team and between the medical team and family should be set as early as possible with milestones set to help guide decisions surrounding futility of ongoing extracorporeal support to meet overall goals of care and quality-of-life expectations.

Bridge to Lung Transplantation

First performed in 1977,⁵⁵ ECMO as a bridge to lung transplantation (LTx) was met with extreme hesitation and ultimately considered by most transplant centers as a contraindication.⁵⁶ However, with improved techniques both in LTx and in ECMO support, case reports and series have demonstrated acceptable outcomes relative to overall LTx survival. The first series describing long-term survival was published in 1993 and described 5 subjects, age 19–46 y, supported with ECMO ranging from 8 to 292 h before LTx, 3 of whom survived > 1 y after LTx.⁵⁷ As LTx experience increased, the field recognized the significant imbalance between organ availability and need with many patients dying while waiting for LTx and an average wait time of nearly 2 y by 1998. In this setting, the need for invasive mechanical ventilator support, let alone ECMO, was felt to be a relative contraindication for LTx as a way to limit organ wastage.⁵⁸

Advances in both transplant and extracorporeal support techniques and understanding continued to push the envelope of therapy. In 2010, a subsequent series and review of both mechanical ventilation and ECMO before LTx described 51 patients included in the United Network for Organ Sharing database from 1987 to 2008.⁵⁹ This summary of the United States experience reported survival at 24 months as 45, 57, and 70% for patients requiring ECMO, invasive mechanical ventilation, or neither before LTx, respectively. Clearly, patients requiring mechanical respiratory support in the form of either mechanical ventilation or ECMO experienced markedly worse survival than patients undergoing LTx with sufficient cardiopulmonary function to breathe spontaneously. In this report, the authors pose the ethical question of whether ECMO as a bridge to LTx was appropriate and how to balance between individual patient benefit and population benefit in the setting of resource scarcity. Nevertheless, in the 2 decades of the 1990s and the 2000s, the number of patients receiving ECMO support simultaneously listed for LTx increased nearly 5-fold from 22 to 104.⁶⁰

As technical sophistication increased in critical care and ECMO management and patients were able to be maintained in an awake state to allow participation in physical conditioning while receiving extracorporeal support, results following ECMO as a bridge to LTx began to close the survival gap. Since 2005, 2 factors have been shown to correlate with improved outcomes with ECMO support before LTx: a mode of respiratory support that allows the ability to ambulate while receiving ECMO and the duration of ECMO before LTx. A report of 11 subjects with a mean age of 34 y that compared the method of respiratory support while receiving ECMO described a 1-y survival of 6 of 7 (85%) in a spontaneously breathing noninvasive support group compared with only 2 of 4 (50%) in a group receiving invasive mechanical ventilation.⁶¹ A review of the United Network for Organ Sharing national database from 2005 to 2013 of adult patients ≥ 18 y old evaluated outcomes based on level of respiratory support at the time of LTx and defined 4 groups: invasive mechanical ventilation only, ECMO only, ECMO with invasive mechanical ventilation, and neither ECMO nor invasive mechanical ventilation. This retrospective database review identified > 12,000 adult LTx recipients, with 796 receiving some combination of ECMO or invasive mechanical ventilation. Patients requiring neither mode of support had the best 1- and 3-y survival at 89.4 and 67.0%, respectively. Patients requiring ECMO alone had lower 1-y survival at 70.4% but equivalent 3-y survival (64.5%) to the no-support group. Need for invasive mechanical ventilation had the worst 3-y outcomes, with only 57.0% survival for invasive mechanical ventilation alone and 45.1% survival for invasive mechanical ventilation and ECMO need at the time of LTx.⁶²

Duration of time receiving ECMO while waiting LTx also appears to impact outcomes, although the data are less robust. In a single report of 17 subjects receiving ECMO bridge who were able to undergo LTx, investigators observed a difference in outcome based on ECMO duration ≤ 14 d (early) or > 14 d (late). The 9 early LTx from ECMO bridge subjects were all alive at 1 y post-LTx compared with only 4 of 8 subjects surviving to 1 y in the late LTx from ECMO bridge group. Notably, the length of invasive mechanical ventilation may have been a significant confounder. The late LTx from ECMO bridge patients received nearly 4-fold longer (45 d vs 12 d, P = .03) invasive mechanical ventilation following LTx than the early group.⁶³

In a subsequent 2015 systematic review, the authors reviewed 14 retrospective clinical case series with a combined total of 441 subjects ranging in age from older teenagers to the 60s and reported 1-y survival between 50 and 90%.⁶⁴ Although these authors found no definitive evidence detailing a best strategy, they concluded that current evidence clearly supported the use of ECMO as a way to sustain select patients with end-stage respiratory failure awaiting LTx. Further, avoidance of invasive mechanical ventilation if at all possible and aggressive pursuit of ambulatory ECMO to preserve physical conditioning appeared to be key in achieving acceptable long-term outcomes and avoiding organ wastage. In all of these reports, subject selection criteria have been crucial. Criteria for eligibility for ECMO support as a bridge to LTx have been proposed and include absence of additional non-pulmonary organ failure, absence of severe neurologic impairment, relatively well-preserved physical conditioning, absence of systemic infection or multidrug-resistant pulmonary infection or colonization, and unmanageable bleeding diathesis.⁶⁵ These exclusion criteria are not absolute and ultimately strive to achieve identification of those patients most likely to survive both an ECMO course and the eventual LTx.

These large case series have described older teenagers and adults. In contrast, the pediatric-specific experience is far more limited. A review of the United Network for Organ Sharing database from 2000 to 2013 included 17 pediatric subjects who were receiving ECMO at the time of LTx. Subsequent comparison with subjects not requiring ECMO at the time of LTx was limited by lack of ability to find an appropriately matched control in 4 and lack of data in another 1, leaving 12 cases available for comparison with matched non-ECMO controls. In this comparison, the authors found no increase in risk of death in the ECMO group with the longest available information on survival up to 4 y after LTx.⁶⁶ Other data on the pediatric experience with ECMO bridge to LTx are limited to case series. In the same paper reporting the United Network for Organ Sharing pediatric experience, the authors describe their own single-center experience of successful ECMO bridge to LTx in 2 of 3 subjects; the third subject died while awaiting transplant.⁶⁶ Another group reported 5 pediatric subjects (age range 10–20 y) with pre-LTx ECMO support undergoing LTx while still receiving ECMO but do not report survival to hospital discharge in this specific sub-cohort.⁶⁷ A separate group report 4 pediatric subjects, age 11–15 y, undergoing ECMO bridge to LTx, all of whom survived to hospital discharge.⁶⁵ The largest individual series described 15 pediatric subjects 0.2–18 y old who underwent ECMO before LTx, only nine of whom were receiving ECMO at the time of LTx. Of the entire group, 6 subjects survived to hospital discharge. There was a trend toward statistical significance for survival in subjects able to wean off of ECMO before LTx (4 of 6) compared with subjects still receiving ECMO (2 of 9) at the time of LTx (P = .09).⁶⁸ Limited to no information was provided in these pediatric-specific case reports on level of activity during ECMO support.

An even smaller number of case reports have described pediatric ambulatory ECMO, demonstrating proof of principle. Thus far, 5 separate case reports have described a total of 9 older children ranging in age from 8 to 19 y who underwent VV-ECMO as a successful bridge to LTx, all but one of whom survived to hospital discharge.^69–73 Most recently, a sixth report described successful use of ambulatory VA-ECMO in a 4-y-old girl who underwent heart-lung transplant for primary pulmonary hypertension and refractory right-ventricular failure and was ultimately discharged home.⁷⁴

Ultimately, ECMO as a bridge to LTx represents a viable and increasingly utilized treatment for what is a terminal condition without LTx. The role of ECMO to either salvage refractory respiratory failure or to prevent complications of invasive mechanical ventilation with subsequent loss of physical conditioning and worsening of ventilator-induced lung injury is only now beginning to be described. Patient selection and thorough and repeated discussions among the multi-professional medical team and family play an essential role during this process. Multiple areas requiring investigation remain, including: (1) timing of ECMO initiation; (2) impact of duration of time receiving ECMO before LTx; (3) relative versus absolute contraindications to use of ECMO as a bridge to LTx; (4) the relative impact of invasive mechanical ventilation versus the ability to maintain an awake and active state even if invasive mechanical ventilation is required; (5) feasibility of this approach in a population where wait times for suitably sized organs are significantly longer than in older teenagers or adults; (6) how best to identify those patients with the highest likelihood of benefit and meaningful survival; (7) the best methods to maintain physical conditioning and rehabilitation in developmentally appropriate but immature children; (8) the health-care costs associated with this approach and resource utilization requirements; (9) ethical considerations regarding the balance between individual versus population benefit; and (10) the informed consent process and how to best achieve a shared understanding of therapeutic limitations balanced with patient-defined goals of care.

Other

ECMO applied to pediatric ARDS and as a bridge to LTx has been most frequently reported in the literature. Use of ECMO for pediatric asthma has limited retrospective data that suggest excellent survival with relatively shorter time spent receiving ECMO but offers little insight into whether ECMO offers morbidity or mortality benefit. A review of the ELSO registry from 1993 to 2007 identified 71 cases of ECMO in pediatric near-fatal asthma with 83% survival to discharge.⁷⁵ In comparison, a review of 261 pediatric ICU subjects with asthma requiring intubation, 3 of whom received ECMO support, demonstrated 84% survival.⁷⁶ A case series of 13 children receiving ECMO for asthma from a single center reported 100% survival with no neurologic sequelae.⁷⁷ However, the question of whether ECMO improves morbidity or mortality when compared prospectively with aggressive medical management remains unanswered.

Single case reports or small case series in children have been published for a multitude of conditions supported with ECMO while the underlying disease was treated either medically or surgically. These include poisonings, life-threatening upper-airway obstruction, trauma, refractory air leak from a multitude of underlying infectious or traumatic causes, and even diffuse alveolar hemorrhage.^47,78–80 As experience builds and ECMO technology and practice continues to advance, the rate of high morbidity or high mortality complications from ECMO support will probably decrease, and more case reports will appear of ECMO use for other indications requiring a brief or prolonged period of cardiopulmonary support pending resolution of the underlying disease process.

Criteria for Separation From ECMO

Even less evidence exists for specific thresholds of lung recovery that allow safe discontinuation from ECMO support. In the absence of unmanageable ECMO complications, current practice principles target demonstration of adequate ventilation and oxygenation without ECMO support before separation and decannulation. In the setting of VV-ECMO, weaning from extracorporeal support is accomplished by decreasing and ultimately discontinuing gas flow (sweep gas) to the oxygenator. Once the sweep gas flow is zero, any oxygenation or ventilation results from the patient's lung function. If the patient is unable to achieve adequate gas exchange, ECMO support can be reinitiated simply by restoring gas flow to the oxygenator. In VA-ECMO, weaning bypass support is achieved by decreasing pump flow and ultimately separating ECMO flow from the patient's circulation while leaving the cannula in place. This can be accomplished by recirculating ECMO flow through a bridge linking the arterial and venous limbs of the circuit with intermittent flashes of flow through the cannula to maintain cannula patency, by disconnecting the circuit and connecting the cannula with tubing and a pump that allows minimal blood flow through an external arteriovenous shunt, or by disconnecting the circuit and instilling a heparin lock into the cannula to prevent clot formation.¹⁰⁰ The time frame over which these weaning maneuvers are accomplished differs in rate and varies from center to center and from individual to individual. The length of time observing tolerance without ECMO support also varies from a few to several hours, depending on the patient's status and initial indications for ECMO.

Specific patient criteria demonstrating stability without ECMO support include markers of adequate oxygenation and ventilation and adequate cardiac output to meet demands. These cardiopulmonary function parameters must then be maintainable with levels of respiratory or cardiac support that provide sufficient room for increased titration in the event of mild to moderate patient decompensation following decannulation. Specific cardiopulmonary function parameters and specific levels of non-ECMO support that correlate with successful discontinuation of ECMO support have not been evaluated. One reasonable approach would be to target OI or oxygen saturation index and alveolar dead-space fraction meaningfully less than the thresholds that triggered initiation of ECMO. Normal lactate and mixed venous oxygen saturation represent cardiac output measures that can also guide decisions regarding ECMO discontinuation.

Ultimately, the underlying disease process prompting ECMO support at the outset must have improved sufficiently that the patient's clinical condition remains stable without ECMO. In those situations where clinical improvement is not achievable and the patient is not an LTx candidate or a catastrophic ECMO complication occurs, making recovery impossible, ECMO should be terminated, and the patient's care goals should be redirected to supportive comfort measures. This decision is clearly difficult and cannot be made unilaterally. Because such a situation is a potential result for every case involving ECMO, attention must be placed on establishing, nurturing, and preserving collaborative relationships among the family, medical, surgical, and ECMO teams from the beginning of ECMO support and through the rest of the ICU and hospital stay. For an introduction to the ethical dilemmas that can ensue in this situation, the last chapter of the ECMO Red Book, 4th edition, provides a concise summary.¹⁰¹

High-Frequency Percussive Ventilation

High-frequency percussive ventilation was first described in the literature in 1988 as a means of achieving the goal of hyperventilation with lower peak airway pressures compared with conventional ventilation (34 cm H₂O vs 62 cm H₂O) in 38 adult trauma subjects with multiple-organ system injury and traumatic brain injury resulting in refractory increased intracranial hypertension.¹⁰² Additional investigators reported positive results with high-frequency percussive ventilation to decrease the incidence of barotrauma and subsequent pneumothorax or subcutaneous emphysema in adult subjects with inhalational lung injury.¹⁰³

In the pediatric population, a subsequent report in 26 children with severe burn injuries and severe inhalational injury found that compared with conventional ventilation, high-frequency percussive ventilation resulted in improved respiratory compliance, lower peak inspiratory pressures, and fewer episodes of pneumonia.¹⁰⁴ Most recently, in a retrospective observational study from Children's Hospital of Philadelphia, 31 pediatric subjects with pediatric ARDS who had failed conventional ventilation demonstrated that within 24 h of high-frequency percussive ventilation initiation, all measures of oxygenation efficiency (OI, oxygen saturation index, P_aO2/F_IO2, and S_pO2/F_IO2) improved significantly with no change in mean airway pressure; ventilation measured by P_aCO2 also improved significantly even as peak airway pressures decreased.⁹³ With the successes seen in their experience with high-frequency percussive ventilation in ARDS, the group in Philadelphia initiated high-frequency percussive ventilation as the standard mode of ventilation for all pediatric patients requiring ECMO support for pediatric ARDS. Compared with historical controls, subjects receiving high-frequency percussive ventilation while receiving ECMO experienced more ECMO-free days at 30 d post-ECMO initiation, indicating both improved survival at 30 d and shorter time spent receiving ECMO in the high-frequency percussive ventilation group. However, subjects receiving high-frequency percussive ventilation during ECMO also received more frequent therapeutic bedside flexible bronchoscopy to assist with secretion clearance.⁹² Whether high-frequency percussive ventilation offers benefit in all patients with pediatric ARDS requiring ECMO support remains to be seen. Certainly, application of high-frequency percussive ventilation, perhaps best applied in conjunction with serial therapeutic bronchoscopies, would fall under a lung recruitment ventilator strategy and appears to be an effective means of clearing inspissated airway secretions.

Pumpless Extracorporeal Gas Exchange

A primary disadvantage of ECMO resides in the pump that drives blood flow through the extracorporeal circuit. Presence of the pump necessarily increases tubing length and exposes red blood cells to physical damage, leading to hemolysis and release of plasma-free hemoglobin, platelet destruction, and an increased inflammatory response. The increased non-organic surface area exposed to blood also serves as an inflammatory trigger. In older patients, a pumpless extracorporeal circuit with an artificial lung (interventional lung assist) provides a low pressure oxygenator that provides supplemental oxygenation and ventilation but depends on the patient's native systemic cardiac function to drive blood flow through the arterial cannula, across the oxygenator membrane, and back to the systemic venous circulation. This external arteriovenous shunt requires relatively intact left-ventricular function and preserved mean arterial pressure in order to maintain sufficient blood flow through the oxygenator. Exact blood flow through the external circuit cannot be controlled externally, and changes in cardiac function or vascular tone greatly compromise the amount of extracorporeal support.¹⁰⁵ In younger pediatric patients, the left ventricle's ability to accommodate the increased output requirements makes this approach less viable.

In contrast, the paracorporeal lung assist device provides an external venoarterial shunt that directs deoxygenated blood from the pulmonary artery through the oxygenator and back into the systemic arterial circulation via the left atrium. Elevated pulmonary artery pressures, as seen in severe pulmonary hypertension, provide the driving pressure gradient that generates blood flow through the external circuit and low-resistance oxygenator. The driving forces here are dependent on increased pulmonary artery pressures. The external circuit offloads the right ventricle in much the same way as an atrial septostomy does with the significant added benefit that the shunted blood becomes oxygenated by the paracorporeal lung assist device. Because of the need for elevated pulmonary artery pressures, this technique has been applied thus far exclusively in patients with primary pulmonary hypertension or pulmonary hypertension related to alveolar capillary dysplasia who were awaiting lung transplantation.^106–108 Each of these patients had initially required rescue with VA-ECMO and was subsequently transitioned to the paracorporeal lung assist device to facilitate improved physical conditioning and separation from invasive mechanical ventilation. The subsequent report detailing the technical aspects and subject experience to date of this approach highlights the ability to minimize sedation, allow extubation, and maximize rehabilitation.¹⁰⁹ Whether this technique can be applied in secondary pulmonary hypertension due to primary parenchymal lung disease is yet to be determined.