ECMO patients and trends

Neonatal extracorporeal membrane oxygenation (ECMO) begins in 1975 with Dr. Bartlett and colleagues at the University of Irvine, CA, USA, where they successfully placed on ECMO Esperanza, a 1-day-old newborn with severe persistent pulmonary hypertension of the newborn (PPHN), after failing conventional therapies [1]. Since then, neonatal ECMO has rapidly evolved due to the increased understanding of cardiopulmonary pathophysiology, advances in medical management, and ECMO technology [2,3,4,5,6,7,8,9,10].

The principle of ECMO can be resumed as follows: the neonate’s deoxygenated blood is taken from the right heart; enters the ECMO circuit; passes through the membrane lung where oxygenation, temperature control, and carbon dioxide clearance are provided; and finally returns to either the arterial system (veno-arterial, VA ECMO, Fig. 1A) or the right heart (veno-venous, VV ECMO, Fig. 1B, C) [11].

Fig. 1
figure 1

Veno-Arterial (VA) ECMO with passive hemofiltration (A); Veno-Venous (VV) ECMO with dual lumen (DL) single bicaval cannula and passive hemofiltration (B); DL bicaval cannula (C)

Full size image

From 1989 to 2020, the Extracorporeal Life Support Organization (ELSO) registry reported 43.707 neonates supported with ECMO worldwide [12].

The number of neonatal respiratory ECMO cases has decreased in the last decade, while cardiac indications remained stable [12]. ECMO survival is variable and depends on the underlying disease [13,14,15,16,17,18,19]. Nevertheless, pulmonary diseases still represent the primary diagnosis for neonatal ECMO (Table 1) [12]. In 2019, 280 newborns (160 respiratory, 93 cardiac, and 27 extracorporeal cardiopulmonary resuscitation (ECPR)) have been supported with ECMO in Europe.

Table 1 Neonatal ECMO: diagnosis and survival rates between ELSO European and international report
Full size table

ECMO criteria in neonates

ECMO may be offered to neonates with severe cardiac and/or respiratory failure, refractory to maximal conventional management, potentially reversible etiology, and high mortality risk [20,21,22,23]. ECMO criteria, as well as contraindications, are reported in Table 2 [22, 24].

Table 2 Neonatal ECMO indications and contraindications (adapted from the ELSO Guidelines for Neonatal Respiratory Failure 2017 [22] and ELSO Guidelines for Pediatric Cardiac Failure 2018 [24])
Full size table

Absolute and relative contraindications are the same for respiratory and cardiac ECMO. Gestational age (< 34 weeks) and low body weight (< 2000 g) have been associated with high rates of intracranial hemorrhage and mortality [20, 25,26,27,28,29,30]. Currently, the neonatal ECMO population is more critically ill than in the past, making patients’ management and cost-benefit assessment more challenging [9, 31]. However, there is wide center-to-center heterogeneity in the lower gestational age limits, based on the local level of expertise [28, 32].

Neonatal respiratory indications

Historically, meconium aspiration syndrome, PPHN, and neonatal respiratory distress syndrome were the most common ECMO indications in the neonatal period. Their incidence has recently declined due to perinatal care improvement [12, 14,15,16,17, 33,33,34,36]. The use of ECMO for pneumonia and neonatal sepsis has also declined. The universal maternal group B streptococcus (GBS) screening and intrapartum antibiotic prophylaxis induced an 80% reduction of GBS neonatal sepsis and, consequently, sepsis-related ECMO use [8, 37, 38]. Unfortunately, ECMO survival for septic shock did not improve [14, 16, 17, 36, 39]. This data could probably be explained by the increase of more aggressive and resistant pathogens, such as Escherichia coli [8, 40].

To date, one-third of ECMO neonatal diagnosis is represented by congenital diaphragmatic hernia (CDH). CDH ECMO has a mean survival rate of 50% that has remained unchanged over time [8, 12, 14, 15, 17, 22, 35, 36]. Although there are reports of a significant improvement in CDH ECMO survival rates in highly specialized centers over the past decade, the overall trend is due to the higher risk profile of the actual CDH population compared to the past [8, 41,41,42,43,44,45,47].

ECMO for congenital diaphragmatic hernia

The pathophysiology of CDH includes pulmonary hypoplasia and PPHN, which are responsible for life-threatening cardiorespiratory failure in the first weeks of life [22]. ECMO has the potential to rescue severe patients, avoiding iatrogenic lung injury, and providing time for improvement of pulmonary hypertension [45, 47,47,49].

Criteria predicting high mortality in CDH patients before ECMO have been recently published based on ELSO data [42, 50], while entry criteria suggested by ELSO are substantially the general ones for respiratory failure and pulmonary hypertension of any cause [22, 48, 51,51,52,54].

In the 5th Edition of the ECLS Red Book [55], CDH ECMO criteria have been integrated with the CDH EURO Consortium Consensus Guidelines Update 2015 (Table 3) [22, 48, 52, 56].

Table 3 ECMO Criteria for CDH patients: a comparison between ELSO and CDH Euro Consortium Consensus indications
Full size table

The optimal mode of ECMO in CDH is still debatable. A systematic review showed no differences in survival rates in the use of VV or VA mode. However, VA ECMO is more extensively used, not only for the advantage of providing both respiratory and cardiac support [57, 58] but also for the lack of adequate double-lumen cannulas for VV ECMO. Timing for optimal diaphragmatic repair in ECMO patients is controversial. Ideally, it is performed when the patient is stable [59, 60]. Some centers do prefer to delay surgery after ECMO weaning to limit the risk of hemorrhagic complications [60,60,62]. However, it has been suggested that early surgical repair while on ECMO may increase survival in the most severe forms of CDH [63,63,65]. Although the optimal timing has not been established, there is an increasing consensus that early repair within the first week of life would improve outcomes [49, 56].

CDH patients usually need long ECMO support, as either the severe vascular disease or pulmonary hypoplasia is less likely to improve over the weeks [8, 12, 42, 49, 66, 67]. Although time limits have not yet been established, a prolonged ECMO course is associated with increased mortality [49].

ECMO for tracheal and bronchial malformations

Congenital tracheobronchial malformations are rare disorders often associated with other intrathoracic anomalies (e.g., left pulmonary artery sling, bronchial extension, congenital heart disease) [68]. Currently, several operative approaches can be used for their treatment, including tracheal resection with end-to-end anastomosis, augmentation tracheoplasty with a pericardial patch, and slide tracheoplasty depending on the type, position, and length of the lesion. Peripheral ECMO (VA or VV) can be considered a valid alternative to cardiopulmonary bypass (CPB) to manage these complex lesions. ECMO requires a lower level of anticoagulation and lasts for several days; further, preoperative ECMO may facilitate the diagnostic workup (cardiac ultrasound, bronchoscopy, and computed tomography with three-dimensional reconstruction), often necessary due to the high prevalence of other intrathoracic anomalies. Intraoperative ECMO may allow surgeons to have better access to the operative field than CPB, which requires a mid sternotomy. Postoperative ECMO may allow lung resting with low-pressure mechanical ventilation favoring the healing of the sutures. The type of ECMO (VA vs. VV) varies among centers and surgical expertise. Despite its possible complications (bleeding, sepsis, neurologic injuries), ECMO may be considered the first extracorporeal approach in unstable neonates with an isolated tracheal lesion. Intraoperative conversion to CPB is used in case of concomitant cardiovascular anomalies [69].

ECMO for neonatal cardiac failure

The use of ECMO for neonatal cardiac failure has progressively increased in the last decade in Europe [13]. Currently, neonatal cardiac ECMO has reported a hospital survival of around 36–55%, which depends mainly on the indications [12, 13, 15, 18, 19]. Pre- and postoperative stabilization of congenital heart disease (CHD) are the most frequent ECMO indication, followed by cardiac arrest, cardiogenic shock, cardiomyopathies and myocarditis, and refractory arrhythmias [12, 13, 35, 70]. Preoperative stabilization (e.g., transposition of great arteries with pulmonary hypertension), failure to wean from cardiopulmonary bypass (stunned myocardium, pulmonary hypertension following cardiac surgery), or postoperative low cardiac output syndrome (Ross-Konno repair, truncus arteriosus repair, anomalous left coronary artery from the pulmonary artery repair, total anomalous pulmonary venous drainage repair) is the most frequent indication for ECMO in neonates with CHD [71, 72]. Risk factors associated with the use of ECMO in CHD are young age, low weight (< 3 kg), pre-ECMO clinical course (mechanical ventilation before surgery > 14 days, presence of fluid overload, cardiopulmonary resuscitation requirements), high complexity of the case according to the STAT classification, presence of shock or arrhythmias, and duration of cardiopulmonary bypass [73].

Cardiac failure associated with septic shock or cardiac arrest (ECPR) is also another ECMO indication in neonates [71]. Hospital survival in these groups is variable and ranges between 40 and 50% [18, 74, 75].

Causes of cardiac arrest and ECPR initiation may impact survival [55]. In the postcardiotomy population, patients with single ventricle physiology received ECMO for cardiac arrest (56%) and low cardiac output state (44%) [76]. ECPR due to hypoxemia or modified Blalock-Taussig shunt thrombosis reported a higher survival than ECPR for low cardiac output syndrome or right ventricular to pulmonary artery shunt thrombosis [71].

Neonatal myocarditis/cardiomyopathy is rare but often requires ECMO to maintain end-organ perfusion [71]. In these cases, ECMO may be used as a bridge to recovery (e.g., viral myocarditis), bridge to heart transplantation or ventricular assist device, or bridge to decision-making. Survival is poor for cardiomyopathies, often fatal, while it is relatively good for viral myocarditis [77, 78].

Arrhythmias (e.g., supraventricular tachycardia, ventricular tachycardia, bradycardia, or severe atrioventricular block) may occur in the perinatal period (primary arrhythmia), postoperatively, or in myocarditis/cardiomyopathy. ECMO support is generally used to maintain end-organ perfusion while optimizing pharmacological/surgical treatment (e.g., catheter ablation, pacemaker implantation) [79].

Timing for ECMO deployment in neonatal cardiac failure is still controversial; however, early initiation may reduce myocardium exposure and peripheral organs to hypoxia [80]. Cannulation for neonatal cardiac failure may be peripheral as well as central according to the circumstances (e.g., failure to wean from cardiopulmonary bypass, the need for high flow). Neonates unable to be weaned from cardiopulmonary bypass are generally supported with central cannulation, while neonates receiving ECPR can receive either a central or peripheral cannulation based on patient clinical history and surgical expertise. Current evidence describing the ECMO cannulation site’s frequency and its impact on outcomes is limited to retrospective single-center studies, mainly focused on ECPR. Further, conflicting results have been reported describing the neurologic complications of carotid versus central cannulation on mortality [81, 82].

ECMO duration is associated with outcome. Prolonged ECMO in children with cardiac disease carries a high mortality and reduces the chances of a successful heart transplant [83]. Survival drops from 45 % (overall survival of children with cardiac disease) to 23–25 % when ECMO duration is between 14 and 28 days and 13% for ECMO course longer than 28 days [71]. Notably, also neonatal respiratory ECMO duration is associated with poor outcome. Prolonged respiratory ECMO > 21 days has reported a survival of 23.5% [9].

Based on all these considerations, when evaluating ECMO as a bridge to a ventricular assist device, its duration should not be longer than 5–7 days [71, 84].

Optimizing end-organ perfusion and oxygen delivery awaiting myocardium recovery is the primary purpose of ECMO in neonatal cardiac failure, and its efficacy is generally reflected by the clearance of acidosis [70]. Thus, when lactate remains high, other causes of hypoperfusion should be appropriately evaluated. Among these, the persistence of residual lesions and the presence of a stunned and dilated myocardium may negatively impact on survival. Postoperative neonatal cardiac surgery patients unable to be weaned off ECMO should be evaluated for residual lesions, and when present, early correction should be advocated [70]. Unloading the left ventricle by creating an atrial shunt via a septostomy or placing another cannula in the left atrium is the most effective strategy to decompress and perfuse a failing left ventricle, especially when a low dose of inotropes is unable to improve ejection [85].

ECMO management

Neonatal ECMO should be provided in a dedicated intensive care unit by a multidisciplinary team, including neonatologists, surgeons, perfusionists, nurses, and psychologists. Psychological support for the family is crucial and should be planned in advance, sometimes also before delivering a special diagnosis (e.g., CDH). A high-quality communication between the ECMO team and the family may also help in ECMO withhold or withdrawal. ECMO management at the bedside starts with the circuit priming and with the choice of the cannulas.

Priming is generally constituted by a mixture of packed red cells and plasma [55]. Electrolytes are added to maintain physiologic levels. Cannulas are generally chosen according to the vessels’ size, often evaluated with ultrasound [86] (Table 4).

Table 4 Characteristics of neonatal cannulas (based on Medtronic Bio-Medicus Pediatric Cannulas, technical sheet 2010; Maquet Avalon Elite Bi-Caval Dual Lumen Catheter, technical sheet 2015) and circuits (based on Permanent Life Support (PLS) System and Quadrox-iD Pediatric (Rotaflow Console with PLS Set))
Full size table

When the cannulas are placed at the neck (either surgical or percutaneously), the position is generally checked with chest X-ray or transthoracic echocardiography [21, 22, 87,87,89]. When ECMO is connected to the patient, the pump flow should be started slowly at 20 ml/kg/min and increased gradually over 5 to 10 min in order to achieve adequate oxygen delivery. For VA ECMO, pump flow is increased to achieve a pre-pump blood oxygen saturation (SpreO2) of 70–80%. On VV ECMO, instead, pump flow is increased to achieve an arterial saturation ≥ 80%.

Although most European centers use centrifugal pumps, differently to the USA, where the roller pump is common practice, this discrepancy in the management would not seem to impact on mortality even if O’Halloran et al. demonstrated contrasting data in favor of the use of roller versus centrifugal pump [90].

As the cardiopulmonary function is gradually replaced, blood gases must be strictly monitored and maintained within normal ranges [22]. SpreO2 is a proxy of tissue oxygenation status, as it reflects the ratio of oxygen delivery to oxygen consumption [22]. Ventilator settings should be tailored at “rest settings” to reduce the risk of VILI, although specific guidelines have not been defined [21, 22]. Serial lung mechanics measurements, such as the forced oscillation technique, may be used to titrate ventilatory support and chest physiotherapy, and support either lung recruitment maneuver and pharmacological treatment [91]. Extubation may be feasible during neonatal ECMO [92].

Fluid management during ECMO is crucial to limit fluid overload and edema [21]. Diuretics, slow continuous ultrafiltration, and continuous renal replacement therapy may be used to achieve dry weight [20, 22, 93]. Full nutritional support should be provided, either enterally or parenterally [22, 94]. Interaction between drugs and non-endothelial surfaces, sequestration into the circuit components, the increase of volume of distribution, and changes in the clearance of many drugs may alter pharmacokinetics and pharmacodynamics; thus, pharmacotherapy should be tailored accordingly [95, 96].

An overview of the critical points of ECMO management is provided in Fig. 2.

Fig. 2
figure 2

Overview of the critical points of ECMO management

Full size image

Anticoagulation

When approaching this critical issue, maturational differences of the hematologic system across age, known as “developmental hemostasis,” should be considered [97,97,98,99,100,102]. Indeed, despite prolonged activated partial thromboplastin time (aPTT) at birth, the neonatal clotting system is balanced and gradually evolves towards a procoagulant phenotype over the first weeks of life [103, 104].

Further, the priming volume (often around three times the neonate’s circulating volume), the use of low ECMO flows, and the contact with foreign surfaces (polyvinyl chloride tubes and polymethylpenthene) may impact hemostasis [105,105,107]. Unfractionated heparin (UFH) is the anticoagulation of choice for neonatal ECMO despite its age-dependent variation in activity [22, 105, 108,108,109,110,112]. As critically ill neonates are also at increased risk of bleeding, namely intracranial hemorrhage, close hemostatic monitoring is required to detect mechanical complications and consumptive coagulopathy [22, 105,105,107, 112].

As each coagulation assay has advantages and limitations, the optimal test is lacking, and instead, a combination of them is usually performed to tailor UFH dosage and blood product replacement (Table 5) [105, 108,108,109,110,111,112,113,114,116]. Besides the use of plasmatic assays, such as aPTT and anti-Xa, many centers adopt whole blood-based tests such as activated clotting time, thromboelastography, or rotational thromboelastography [11, 22, 105, 109, 110, 117,117,118,119,120,121,123].

Table 5 Assessment of coagulation status and management during neonatal ECMO
Full size table

Although UFH has molecule- and age-dependent variation in activity, it is the anticoagulant of choice [22, 105, 108,108,109,110,112].

Normal antithrombin levels are essential for UFH efficacy; however, they are physiologically lower in neonates than in children and adults. Evidence for antithrombin supplementation has yet to be proven, and there are increasing concerns regarding thromboembolic events associated with AT supplementation, especially in critically ill newborns and children [124,124,125,126,128]. The presence of high levels of latent antithrombin, with potent pro-coagulative effects, in the commercially available products could explain this phenomenon [129].

Direct thrombin inhibitors are a promising alternative to UFH for anticoagulation in neonates, especially in heparin-induced thrombocytopenia, heparin resistance, or significant thrombosis. Dosing strategies and risk of adverse effects are still a matter of debate [22, 101, 105, 110, 130,130,131,132,134].

Overall, there is still a wide heterogeneity in coagulation management during neonatal ECMO across centers [109, 113, 135,135,136,137,138,139,141].

Weaning, decannulation, and withdrawal

As minimal support is reached to ensure tissue oxygenation, decannulation is considered. Trial-off support is performed to test the patient’s readiness for weaning before any attempt of decannulation [22].

During VV ECMO, native gas exchange is evaluated by progressively reducing sweep gas flow while adjusting ventilatory support. When the gas flow is stopped and gas exchange is adequate with minimal ventilator support, patients are ready to be weaned. During VA ECMO, the blood flow rate and gas flow is gradually reduced over time while adapting the ventilator setting and monitoring patients’ status. Alternatively, a bridge may be inserted into the circuit between the venous and the arterial side, to allow blood flow to re-circulate within the circuit, thus excluding the patients. Achieving low flow conditions in neonates is often impossible due to the required minimum flow for the ECMO devices (100–200 ml/min). Thus, other groups successfully promoted trial-off with a retrograde pump flow of the ECMO circuit [142]. If trial-off is tolerated, the patient is ready for decannulation [21, 22]. The carotid artery’s repair is rarely performed and is still a matter of debate [22, 143].

ECMO should also be discontinued if there is no hope of survival, there is no chance of organ replacement, or the patient has not shown any improvement during a reasonable amount of time, which must be defined on a case by case basis [21].

Complications

Critical issues related to hemostatic disturbances make hemorrhagic and thrombotic complications prevalent, with decreased survival rates [144, 145]. Circuit clotting is the most frequent mechanical complication, being the oxygenator the most common site. Coagulation abnormalities may lead to ischemic and hemorrhagic injuries [12, 22]. Since brain injuries are worrisome during neonatal ECMO, bedside imaging and neuromonitoring are mandatory during and after ECMO [22, 146,146,148].

Among the most severe neurological complications, seizures, ischemic infarction, and brain death are frequent and associated with poor prognosis [149, 150]. Cerebral function monitoring, continuous electroencephalography, and near-infrared spectroscopy may help to detect neurologic events, although pediatric literature is still limited [11, 151].

A degree of renal impairment, ranging from oliguria to renal failure, frequently occurs while on ECMO. Acute kidney injury and fluid overload are independent risk factors associated with increased mortality [152, 153]. Chronic kidney injury is less frequent since the long-term renal function is usually restored [154].

Cannulas, central lines, immobility, and general conditions are risk factors for infections associated with increased ECMO duration and higher mortality [20, 155]. Routine prophylaxis is not recommended unless a specific culture demonstrates an ongoing infection, while antibiotic coverage for cannulation should follow standard surgical prophylaxis [156,156,157,159].

Arrhythmias and myocardial dysfunction are common in VA mode. Other complications, like cannula displacement and cardiac tamponade, are life-threatening and need to be promptly recognized and corrected (Fig. 3) [20].

Fig. 3
figure 3

Complications during neonatal respiratory and cardiac ECMO and incidence

Full size image

Long-term morbidities and follow-up

Improved survival carries a higher risk of long-term morbidity among survivors depending on several factors, such as the primary underlying condition and adverse events arising during the ECMO course [12, 13, 49, 160,160,161,162,163,164,165,167].

Among respiratory ECMO patients, the risk of developmental disabilities is significantly increased [150, 168,168,169,170,172]. Bilateral sensorineural hearing loss, motor problems, learning difficulties, behavior disorders, and cognitive and neuropsychological impairments require specific assessment and multidisciplinary long-term follow-up [160, 161, 166, 167, 173,173,174,175,176,178].

Nevertheless, adult ECMO survivors show high self-esteem and are satisfied with their quality of life [179,179,181].

The need for oxygen supplementation, pulmonary hypertension, obstructive pattern with bronchospasm, asthma, decreased exercise tolerance, and chronic lung disease are among the most common conditions within the long-term respiratory morbidity [59, 161, 162, 173, 182,182,183,185].

The growth pattern is generally typical or slightly decreased in ECMO survivors, with good chances of catch-up growth during childhood and adolescence [162, 173, 186].

Most deaths occur while on bypass or early on in the hospital course, while late deaths occurring after ECMO discontinuation are uncommon and mainly affecting CDH patients (14.3%) [187].

There are considerable data regarding survival to hospital discharge of neonates receiving ECMO for cardiac failure; however, limited data exist on intermediate- and long-term outcomes. Among survivors, 40–60% reported new neurologic deficits and low health-related quality of life than healthy controls [188,188,190]. Neurologic complications may have variable features (language acquisition delay, motor development stages delay, behavioral problems) and represent a source of significant morbidity and long-term care requirements [190]. Due to the cohorts’ heterogeneity, which includes different age groups and different ECMO indications (ECPR, low cardiac output syndrome, septic shock, myocarditis), a robust conclusion cannot be achieved.

In the light of these considerations, standardized follow-up protocols among centers and data sharing are important to provide further knowledge and implement interventions to prevent and manage complications, improve quality assistance, and improve the overall quality of life.

Conclusions and future directions

Over the years, ECLS technology and expertise have considerably improved, pushing the lower limit of gestational age and suggesting that ECMO use may be potentially extended to more premature infants in the next years [28].

Experimental models of artificial placenta have been developed [191,191,193]. An ideal artificial placenta provides ECMO through the umbilical vessels, thus preserving major fetal blood vessels from cannulation. It uses low partial pressure of oxygen because of fetal hemoglobin features and increases hematocrit levels. No ventilation is required since the infant could “breathe” with fluid-filled lungs (perfluorocarbon liquid ventilation), reducing the risk of VILI to the developing lungs [194,194,195,197]. A trial using an artificial placenta in preterm lambs showed increased survival compared to mechanical-ventilated lambs [192]. An experimental extra-uterine system to physiologically support fetal lambs has shown encouraging results in terms of hemodynamic stability, oxygenation, and maintenance of fetal circulation [193].

Advances in technology, materials, and miniaturization of equipment would help to reduce ECMO-related complications. Biomimetic or biopassive tubing may mitigate the adverse effects of the contact between blood and non-endothelial surfaces, but their efficacy is partial [101]. The endothelization of the circuit would provide physiological regulation of coagulation and inflammatory response in ECMO patients. However, the in vitro process is technically challenging, takes time, and cannot be produced in urgent settings [198,198,199,201]; the in vivo process is currently not feasible, but ongoing research is promising. Lastly, paracorporeal lung assist devices might be explored in selected cases where traditional ECMO is not an option, although evidence for hybrid approaches is scarce [202].