Neurological Monitoring and Complications of Pediatric Extracorporeal Membrane Oxygenation Support

Abstract

Extracorporeal membrane oxygenation (ECMO) is extracorporeal life support for life-threatening cardiopulmonary failure. Since its introduction, ECMO use has expanded to patients with more complex co-morbidities without change in patient mortality rates. Although many patients survive, significant neurologic complications like seizures, ischemic strokes and intracranial hemorrhage can occur during ECMO care. The risks of these complications often add to the complexity of decision making surrounding ECMO support. In this review, we discuss the pathophysiology and incidence of neurological complications in children supported on ECMO, factors influencing the incidence of these complications, commonly used neurologic monitoring modalities, and outcomes for this complex patient population. We discuss the current literature on the use of electroencephalography (EEG) for both seizure detection and monitoring of background EEG changes, in addition to the use of less commonly used imaging modalities like transcranial doppler (TCD). We present the knowledge gap and the lack of clinical concensus guidelines in managing these potentially life changing neurological complications. Additionally, we present future directions to further understand the pathophysiology of ECMO-related neurological complications.

Introduction:

Extracorporeal membrane oxygenation (ECMO) is an advanced life support technique employed in intensive care units to support patients with cardiopulmonary failure. Venous blood is drained from the body into the extracorporeal circuit and pumped through an oxygenator for gas exchange. Oxygenated blood is then returned to the patient either through the venous circulation in venovenous ECMO (VV ECMO) or the arterial circulation in venoarterial ECMO (VA ECMO). ECMO has grown since its inception to become an invaluable tool in supporting children and adults with severe cardiac and pulmonary failure refractory to conventional management¹. ECMO utilization has quadrupled in children and increased ten-fold in adults in the last two decades^2,3. However, patients supported on ECMO continue to be at significant risk for neurologic injury, including hypoxic ischemic injury in the peri-cannulation period, as well as thromboembolic stroke and intracranial hemorrhage while on ECMO support. Neurologic injury on ECMO is associated with increased risk of mortality and neurologic disability among survivors⁴. In this review we briefly discuss the history of ECMO support in children, neurologic complications associated with extracorporeal life support, and future directions in neuromonitoring of ECMO patients, focused on minimizing neurologic complications and their sequelae in this vulnerable population.

Pediatric ECMO and pathophysiology of neurologic injury during ECMO

The concept of artificial oxygenation and perfusion was first used successfully for cardiac surgery in 1953⁵, and the first successful long-term support for severe respiratory failure in an adult patient with post-traumatic acute respiratory failure was reported in 1972⁶. The first report of successful use of ECMO in a newborn with severe respiratory failure by Bartlett in 1976⁷ led to rapid expansion of extracorporeal support in children, particularly newborn infants, over the following two decades⁸.

The cannulation strategy and circuit configuration depend on the type of support needed: cardiovascular (VA ECMO), respiratory (VV ECMO) or both (VA ECMO), as well as the clinical context (e.g., emergent extracorporeal cardiopulmonary resuscitation [ECPR] with percutaneous bedside cannulation, ECMO initiation for cardiogenic shock post-cardiac surgery with central cannulation via sternotomy, etc.). ECMO involves cannulation of the heart or of one or two large vessels, typically the right common carotid artery and the right internal jugular vein in infants and small children, and the femoral vessels in older children and adults ^9,10,11. Some ECMO centers place cephalad jugular catheters in infants with a jugular venous ECMO cannula, in an effort to augment venous return, reduce atrial recirculation, and decompress the cerebral venous circulation. The most recent evaluations of this practice showed similar outcomes, including survival and neurologic complications, when comparing ECMO patients with vs without cephalad jugular catheter placement ¹².

Rates of neurologic complications are higher in infants and children supported on VA ECMO compared to those supported on VV ECMO ¹³. While VV vs VA cannulation decisions are made driven primarily by type of support needed, the added physiologic impact of a carotid or aortic cannulation are nontrivial. With an intact circle of Willis, blood supply can reach the right hemisphere normally perfused by the ligated carotid artery. However, blood flow to the right hemisphere may have decreased velocity, particularly with concomitant decreased venous outflow from the ligated right internal jugular vein ^14–16. ECMO also appears to alter cerebral blood flow within smaller vessels. Animal studies demonstrate impaired cerebrovascular autoregulation at lower perfusion pressures ¹⁷, likely due to lack of flow pulsatility or exhaustion of compensatory dilation with lower flow states ¹⁸. The effects of carotid artery cannulation on long-term neurologic outcomes remain unclear, especially as there is variability in practice in terms of the carotid artery repair at the time of decannulation, varying from 6.6% of all patients to 10.7% of patients older than 5 years ^19–22.

The alteration of cerebral blood flow and cerebrovascular autoregulation during ECMO, combined with the underlying pathophysiology of critical illness and the use of sedative medications, contribute to decreased cerebral metabolism during ECMO. This is reflected in reduced power and amplitude on electroencephalography (EEG), and a global decrease in power on quantitative EEG during ECMO, which can be restored after decannulation from ECMO ²³.

Lastly, exposure to the foreign surfaces of the ECMO circuit triggers inflammatory and pro-thrombotic responses ^24–27 that are additive to the patients’ critical illness processes such as disseminated intravascular coagulopathy [DIC] ²⁸, sepsis, autoimmune diseases, pregnancy or trauma ²⁹. To avoid clot formation in the circuit, ECMO patients are typically anticoagulated using a continuous infusion of unfractionated heparin or a direct thrombin inhibitor. Patients are therefore at risk for both thromboembolic events due to clots dislodged from the circuit, and hemorrhagic events associated with long-term anticoagulation in the presence of DIC, surgical wounds, etc. Bleeding (intracranial hemorrhage [ICH]) and thrombotic (stroke) neurologic complications can occur at any time during the days, weeks or months of ECMO support ²⁶.

Outcomes of critically ill children supported on ECMO

Severe refractory cardiopulmonary failure or cardiac arrest requiring ECMO support in children continues to be associated with high mortality. In general, survival trends have been stable over the last decade. In a recent report of Extracorporeal Life Support Organization (ELSO) registry data, neonatal respiratory, pediatric respiratory and pediatric cardiac ECMO survival rates were around 61%, while neonatal cardiac, and neonatal and pediatric ECPR survival rates were within three percentage points of 42% ³.

A systematic review of the literature of neurologic outcomes following ECMO support in infants and children summarized data from 60 articles published between 2000 and 2016 ³⁰. At hospital discharge, 48% of survivors (cumulative n=673) had favorable outcome defined as Pediatric Cerebral Performance Category (PCPC) <3 or unchanged from baseline. Long-term follow-up showed that 10% to 50% of children scored more than 2SD below the population mean on cognitive testing, 16% to 46% displayed behavioral problems, and 12% had evidence of severe motor impairment ³⁰. The wide ranges in disability rates reported were due to heterogeneity of underlying diagnoses, of outcomes measures employed, and varying age at follow-up ³⁰. Overall, neurologic outcomes in congenital diaphragmatic hernia (CDH), cardiac disease, and cardiac arrest populations tended to be worse than those in respiratory or mixed ECMO populations ³⁰.

New data are emerging from more reports published since 2016, with mixed results. In a mixed neonatal and pediatric ECMO population, results from a multicenter U.S. cohort study showed that, at hospital discharge, 32% of 282 survivors had good, 40% mildly abnormal, 24% moderately abnormal, and 5% severely or very severely abnormal function based on the Functional Status Scale ³¹. A standardized 1-year follow-up clinic after respiratory ECMO identified neurodevelopmental problems in 30% survivors who attended the clinic (n=98) ³². In a two-center U.S. cohort study conducted in a mixed neonatal and pediatric ECMO population, structured 6-month and 1-year post-ECMO neurologic and neuropsychological evaluations showed that ECMO survivors had scores for adaptive behavior, cognitive, neurologic and quality of life that were all below the general population means, but mostly considered minor within each domain ⁴. The median Vineland Adaptive Behavior Scales-II (VABS-II) score was 91 (interquartile range [IQR], 81-98) (general population mean, 100; SD, 15), with 65% of survivors who attended the study follow-up visits having a VABS-II score ≥85 (n=40) [4]. Lastly, in a secondary analysis of the Therapeutic Hypothermia After Pediatric Cardiac Arrest (THAPCA) trials, 30.5% of study participants who underwent ECPR survived to 12 months with VABS-II ≥70 ³³. While the heterogeneity of patient populations and outcomes measures used precludes aggregate analyses, in these recent cohorts, approximately one third of surviors to hospital discharge or long-term follow-up display moderate to severe neurodevelopmental, behavioral, or functional deficits.

Neurologic complications in pediatric ECMO patients

Neurologic injury associated with ECMO may be sustained prior to, during, or after ECMO support ³⁴. Despite the increase in ECMO use over the last decade, the incidence of reported neurological complications has not changed significantly³. Table 1 summarizes some of the reported ECMO related risk factors associated with neurological complications ^{21,31,35–41}.

Table 1.

Reported risk factors associated with neurologic complications and post-discharge functional, neurobehavioral or cognitive deficits in children requiring ECMO.

Factor potentially increasing risk	Neurologic complication					Reference
	Seizure	Stroke	Any CNS complication	Brain Death	Post-discharge deficits
ECMO Indication
ECPR		ⅹ	Ⅹ			31,35–37,41
ECMO modality and cannulation site
VA			Ⅹ			31,35,37
Carotid		Ⅹ	Ⅹ			21,36,38
Femoral				Ⅹ		36
Patient factors
Younger age	Ⅹ	ⅹ	Ⅹ			37,39–41
Renal Failure		ⅹ			Ⅹ	31,35
Hepatic dysfunction					Ⅹ	31
Sepsis		ⅹ				41
Pre-ECMO chronic conditions					Ⅹ	31,33
Longer hospitalization					Ⅹ	31
Lactate					Ⅹ	33
Creatinine >3.0 mg/dL			Ⅹ			35
pH <7.2 pH <7.1	Ⅹ	ⅹ	Ⅹ Ⅹ			35,40,41 37
Inotropic support		ⅹ	Ⅹ			35,41
Thrombocytopenia <30 cells/µL	Ⅹ					40

ECMO = extracorporeal membrane oxygenation, ECPR = extracorporeal cardiopulmonary resuscitation, VA = venoarterial, RBC = red blood cells

Commonly reported neurological complications include seizures, ischemic strokes and intracranial hemorrhage:

Seizures

Several studies have shown that seizures occur in approximately 20% (18%−23%) of neonates and children supported on ECMO ^40,42,43 (Table 2).

Table 2.

Reported incidence of seizures in children on ECMO support monitored by continuous EEG

Year	Author	n	Agea	% ES	% NCS	% SE
2013	Piantino et al.	19	2.7 (0.5-6.8) years	21%	16%	10%
2017	Lin et al.	99	20 (3-120)b vs 13 (3-364)c days	18%	15%	11%
2018	Okochi et al.	70	0.6 (0.2-3.6)b vs 6.7 (1.1-62.3)c months	23%	12%	5%

n, number of patients; ES, electrographic-confirmed seizures; NCS, non-convulsive seizures; SE, status epilepticus

^aAge reported as median (interquartile range);

^bSeizure group,

^cNo seizure group

The incidence of electrographic seizures without clinical correlates ranged between 12% and 16% of all monitored patients, with evidence of status epilepticus in over half of these patients. The majority of the identified seizures occurred in the first 24 hours of monitoring following initiation of ECMO support ^40,42,43. The incidence of seizures correlated with the presence of intracranial pathology on further head imaging and with higher mortality and unfavorable neurodevelopmental outcomes ⁴⁴, findings consistent with others evaluating the impact of seizures on long term functional outcomes in children ^45,46.

Similar rates of seizures, convulsive and non-convulsive, have been reported in ECMO patients monitored with intermittent EEG ⁴⁷. The July 2019 ELSO registry international report showed EEG-confirmed seizure rates ranging between 1.9% in pediatric patients on respiratory ECMO support to 11.4% in neonates who underwent ECPR (data for 2018 ELSO registry entries) (Table 3) ⁴⁸. The noticeably lower seizure rates in the ELSO registry are likely an underestimate of the true prevalence of seizures during ECMO, as detected by single center quality improvement or research studies deploying active EEG monitoring.

Table 3.

Neurological complication rates by age group and mode of ECMO support for 2018 as reported in the July 2019 ELSO registry international report⁴⁸

	Neonatal	Pediatric
Seizures confirmed by EEG
Respiratory ECMO	4.8%	1.9%
Cardiac ECMO	5.4%	3%
ECPR	11.4%	8.3%
Central nervous system bleeding confirmed by ultrasound, CT or MRI
Respiratory ECMO	10.1%	7.8%
Cardiac ECMO	11.1%	5.1%
ECPR	17.7%	8.1%
Central nervous system infarction confirmed by ultrasound, CT or MRI
Respiratory ECMO	3.4%	5%
Cardiac ECMO	2%	3.1%
ECPR	4.4%	7.2%

The etiology of seizures in neonates and children supported with ECMO is multifactorial. Risk factors include hypoxic ischemic injury following cardiac arrest, and presence of intracranial hemorrhage or ischemic stroke⁴⁵. Earlier attempts to link the pathophysiology of seizures to alterations in cerebral blood flow during ECMO by examining patients with subclavian steal phenomenon in the setting of carotid artery cannulation were not fruitful ⁴⁹.

Seizures increase a child’s risk for unfavorable short-term neurologic outcomes ⁵⁰, death ^51–53, and increased rates of cerebral palsy and cognitive deficits at long-term follow-up^44,54. The frequency and potential long-term impact of seizures emphasize the need for vigilant EEG monitoring, with prompt detection and treatment of seizures for all patients supported on ECMO.

Ischemic stroke

Ischemic strokes occur in up to 33% of pediatric ECMO patients ^38,55,56, with much lower rates reported in the ELSO registry (2-7.2%) (Table 3). In children, strokes have a variety of presentations, including a mix of both focal and diffuse signs (e.g., altered mental status) in half of cases, and seizures in up to a third of all cases ⁵⁷. Standard medical care for ECMO, including the use of sedation, potential need for neuromuscular blockade and post-cannulation patient positioning may render detection of early stroke signs challenging. Risk factors for stroke during ECMO include circuit thrombosis with dislodgement of emboli, and acute alterations in cerebral arterial circulation with carotid artery cannulation, although studies are mixed as to whether carotid cannulation increases ischemic stroke risk ²¹ or not ⁵⁸.

Intracranial hemorrhage (ICH)

Despite advances in monitoring of hemostasis and anticoagulation management on ECMO, bleeding events complicate up to 66%-80% of ECMO runs ^48,59. Incidence of ICH in the literature ranges between 3.6% in combined adult and pediatric registry studies and up to 16% in pediatric-only cohorts ^59–61. This rather wide range may be explained by underdetection and underreporting in registry data and the lack of standardized screening and uniform definitions for ICH. Reported ICH diagnoses include large intraparenchymal hemorrhages (IPH), petechial IPH, subarachnoid hemorrhage, subdural hemorrhage and combinations thereof ^62,63. Subtle abnormalities such as cerebral microbleeds have also been detected on post-ECMO MRI, with unclear clinical significance at this time and phenotypes ranging from asymptomatic to persistent coma ^64–68. The incidence of ICH appears to be highest within the first 4 to 5 days of the ECMO course ^59,69. ICH has been shown to correlate with both increased mortality and worse functional neurologic outcomes ^59,70.

In adults, risk factors for ICH on ECMO include disease-related factors such as pre-ECMO cardiac arrest ⁷¹, sepsis ^41,60, , influenza⁷¹, renal failure ⁶³, use of renal replacement therapy, pre-ECMO duration of mechanical ventilation and low admission fibrinogen ⁶³. ECMO-related risk factors include thrombocytopenia ⁶³, hemolysis ⁷¹, inotropic support ⁴¹ and acute rapid increase in PaO₂ and decrease in PaCO₂ with the initiation of VV ECMO ⁶⁸. Cerebral venous congestion due to obstruction of venous outflow from cannulation of the right internal jugular vein may also increase the risk of ICH ⁷². Interestingly, in a single center study, patient coagulation profiles were not predictive for acute ICH or infarct on ECMO ⁷³.

Neurologic monitoring of ECMO patients

The combination of critical illness with hypoxia, acidosis, and multiple organ failure, cannulation and ligation of major neck vessels, impaired cerebrovascular autoregulation, and concomitant risk for thrombus dislodgment from the ECMO circuit and bleeding due to exposure to systemic anticoagulation, render ECMO patients at high risk for neurologic complications. Clinical monitoring by bedside examination is often limited by sedatives and possibly neuromuscular blockade. Additional neuromonitoring methods therefore enhance neurologic assessment during ECMO and assist with early detection of neurologic injury ⁷⁴. Table 4 summarizes the recommended guidelines for pediatric and neonatal neuromonitoring during ECMO support ^75,76.

Table 4.

Neuromonitoring recommendations for pediatric and neonatal patients on ECMO support

Age group	ECMO indication	Recommendation
Neonatal	Respiratory failure	• Serial cranial ultrasound while on ECMO • Brain MRI prior to discharge • Continuous EEG monitoring for at least 24 hours
Pediatric	Respiratory failure	• Head CT or brain MRI prior to discharge • Continuous EEG monitoring for at least 24 hours
Neonatal Pediatric	Cardiac failure	• Screening cranial ultrasound and more definitive neuroimaging (head CT) as needed during ECMO • Brain MRI prior to discharge • Continuous EEG monitoring for at least 24 hours

ECMO = extracorporeal membrane oxygenation, MRI = magnetic resonance imaging, CT = computed tomography, EEG = electroencephalography

A. Modalities to monitor brain physiology that can impact ECMO management

Electroencephalography (EEG)

EEG is used to assess for the presence of seizures and to evaluate the brain’s background electrical activity and organization, including state change, asymmetry, and focal abnormalities such as slowing or epileptiform discharges (Figure 1). EEG background abnormalities are associated with mortality and unfavorable neurodevelopmental outcomes in neonatal and pediatric patients ^51,77,78,79. EEG monitoring requires experience in both set-up and interpretation but has become the recommended modality of monitoring for critically ill children at risk for seizures and acute neurologic injury, including those supported on ECMO ⁷⁵. The American Clinical Neurophysiology Society (ACNS) Consensus Statement on continuous EEG (cEEG) in Critically Ill Adults and Children designated children on ECMO requiring sedation and paralysis as a high-risk group that should be monitored by cEEG ⁷⁵.

Figure 1: Examples of EEG tracings for patients supported on ECMO.

A. Raw EEG tracing of a 7 year old with respiratory failure requiring VA ECMO support. Continuous video EEG monitoring initiated soon after cannulation revealed status epilepticus (black arrows on raw EEG tracing and density spectral array) and subsequent seizures during recording. B. Raw EEG (1 and 2) and asymmetry index from quantitative EEG (3) of a 1 year-old child on VA ECMO support after near-drowning. While unplugged from EEG recording for head CT the patient suffered a right hemispheric stroke. (1) Raw EEG tracing with symmetric sleep spindles corresponding to normal asymmetry index. Upon return from CT suite, the patient demonstrated asymmetry of sleep spindles (2) and dramatically increased relative power or activity (dark blue in 3) on the left and not right hemisphere.

There are recognized challenges of EEG monitoring. Placement of EEG electrodes for extended periods of time poses the risk of compromising scalp integrity, especially in neonates. Electrical artifact from the ECMO circuit can mimic pathologic EEG findings, particularly in younger children ⁷⁴. However, these challenges are easily mitigate with experience, expertise, and careful skin monitoring. There is growing interest in the utiity of bedside real time compressed EEG data as Digital Spectral Array (DSA) to aid clinicians in early seizure detection ^80,81. Prompt recognition and quantification of seizures is important as there is growing evidence of the correlation between seizure burden and worse brain injury and outcomes ⁸².

Imaging

Ten to 60% of pediatric ECMO survivors have been found to have abnormal neuroimaging studies ^83–85. Most ECMO centers have institutional clinical protocols for the type and timing of neuroimaging studies employed during and following ECMO (cranial ultrasound, brain CT or brain MRI) explaining the wide range of these reported incidences.

Cranial Ultrasound

Ultrasonography of the brain is a non-invasive, portable imaging option for infants with an open fontanelle. Ultrasound has become part of the routine clinical care of infants supported on ECMO. ELSO guidelines recommend obtaining a pre-ECMO ultrasound followed by daily head ultrasounds for the first week on ECMO, then every-other-day screening or more frequent if indicated ⁷⁶. Raets et al proposed a schematic classification of neuroimaging abnormalities detected by ultrasound in neonatal ECMO consisting of vessel occlusion (i.e., arterial ischemic stroke, venous sinus thrombosis), primary hemorrhage (i.e., extracerebral, choroid plexus, parenchymal) and other abnormalities (e.g., postasphyxial injury, preterm white matter injury, etc) ⁶¹. Abnormal ultrasound findings have not been shown to predict neurodevelopmental outcomes ^86,87, although the type and location of certain lesions detected by ultrasound may correlate with neurologic outcomes. Cranial ultrasound is limited by the size of an open anterior fontanelle, limited field of view, and diagnostic variability ⁸⁸. In a single center study of 50 neonates with a brain MRI after ECMO, 50% of infants with normal cranial ultrasounds during ECMO had an abnormal MRI⁸³. For these reasons, cranial ultrasound remains a screening tool, with brain CT and MRI recommended as confirmatory studies⁷⁶.

Measurement of resistive indeces (RI) using cranial ultrasounds represents a potential neurologic monitoring method for infants on ECMO ⁸⁹. The variability of the RI on Doppler sonography measured at baseline and after gentle fontanel compression has been proposed as a test to detect early disturbances in cerebrovascular regulation. In a single center study, RI variability in the anterior cerebral artery of 10% or greater on any ECMO day was associated with increased risk for cerebrovascular complications in neonates supported on ECMO⁸⁹.

Transcranial Doppler (TCD)

TCD is the application of ultrasound technology to sonate blood flow through major arteries. Unlike cranial imaging ultrasounds, TCD is not limited to children with open fontanelles, and allows for real-time monitoring of cerebral blood flow patterns in the broader pediatric critical care population ⁹⁰. Flow patterns for children supported on ECMO are altered compared to age-matched normal controls ⁹¹. Higher than normal cerebral blood flow velocities on ECMO may signal cerebral hemorrhage days prior to clinical recognition ⁹². Increased pulsatility index may also correlate with ischemic injury ⁹³. As with cranial ultrasound, wide use of TCD is rendered difficult by limited fields of view and operator dependence ⁸⁸.

Computed Tomography (CT)

Older infants and children cannot be monitored with serial cranial ultrasounds while on ECMO, thus neuroimaging studies during ECMO in this patient population are limited to head CT. Head CT is mainly performed when there is clinical suspicion of acute neurologic injury (Figure 2). Its use is limited by the need for transport (unless portable CT is available)⁹⁴, exposure to radiation and relative insensitivity to early ischemic injury. Recent studies have shown higher detection rates of ischemic stroke and ICH with more active CT use, reflecting the possible underreporting in centers with less frequent imaging ^55,69. Advantages of CT include a full field of view of the brain, operator independence and excellent detection of ICH.

Figure 2.

9 month old infant on VA ECMO support for refractory septic shock and multiorgan dysfunction. Within first 24 hours of ECMO course, the patient developed clinical signs of right sided facial venous congestion and asymmetric slowing on EEG, initially thought to be secondary to superior vena cava obstruction by the ECMO venous drainage cannula. A. Head CT on ECMO day 3 demonstrated multifocal ischemic infarcts within the right hemisphere for which his systemic anticoagulation was temporarily held. B. Follow up head CT on ECMO day 5 demonstrated evolution of the ischemic infarcts and hemorrhagic conversion within the posterior ischemic infarct. He was decannulated on ECMO day 9 and discharged home on hospital day 68 with residual left hemiparesis.

Detection of neuroimaging abnormalities, ICH or ischemic stroke, may affect anticoagulation strategies and the ability to continue extracorporeal support and systemic anticoagulation, depending on the severity of the recognized complication.

Near Infrared Spectroscopy (NIRS)

NIRS technology utilizes light in the near-infrared spectrum to distinguish between oxygenated and de-oxygenated red blood cells, reflecting changes in tissue oxygenation and blood volume ⁹⁵. This technology has been present for the past two decades, but its clinical applications are still expanding. Absolute or relative (e.g., decline by >20% from baseline) reduction in regional cerebral oximetry has been shown to be associated with neurological injury confirmed by imaging in critically ill children ^96–98 and adults ⁹⁹ supported on ECMO.

B. Monitoring modalities for outcome prognostication

Plasma brain injury biomarkers

Recent years have seen an increasing number of studies investigating plasma brain injury biomarkers as both neuromonitoring and prognostication tools in neurocritical care conditions, spearheaded by initial efforts in the field of traumatic brain injury ¹⁰⁰. The complexity of brain injury in ECMO patients is unlikely to be captured by a single biomarker, but panels of biomarkers may prove useful in distinguishing different pathologic-anatomic processes. High concentrations of brain-specific proteins circulating in blood have been shown to be associated with abnormal neuroimaging findings during or immediately following ECMO support, with survival after critical illness requiring ECMO support, or with neurofunctional status at hospital discharge in former ECMO patients^101–106. These proteins reflect neuronal injury (e.g., neuron specific enolase [NSE]), astrocytic injury (e.g., glial fibrillary acidic protein [GFAP], S100b), and neuroinflammation (e.g., monocyte chemoattractant protein 1 [MCP1], intercellular adhesion molecule-5 [ICAM-5])^101–106.

Magnetic resonance imaging (MRI)

ECMO equipment is not MRI compatible and brain MRI studies are therefore limited to the post-decannulation period. As such, MRI is not technically a neuromonitoring modality while on ECMO. MRI is more sensitive and specific for intracranial pathology compared to cranial ultrasound and head CT and does not involve expose to radiation ^83,85. However, there is institutional variability in indications for and timing of post-ECMO MRI ⁷⁴. The current ELSO guidelines recommend neuroimaging in the form of head CT or MRI prior to discharge in neonatal and pediatric patients supported on ECMO for any indication ⁷⁶.

Future directions

The contemporary ICU allows for multimodal neuromonitoring of critically ill patients at risk for acute neurologic injury. Increasingly, patients are monitored and systems are developed for concomitant cerebral oximetry, continuous EEG, non-invasive cerebrovascular autoregulation monitoring and serial TCD monitoring ^74,107–109. Quantitative EEG or cerebrovascular autoregulation software allowing for rapid interpretation of otherwise complicated and difficult-to-interpret continuous data are being evaluated in at-risk ICU populations ^110–114. Emerging plasma biomarkers of brain injury could provide a real-time, non-invasive measure of acute neurologic injury and recovery ^101,103,115, with promising progress towards the development of point-of-care devices to measure such biomarkers ¹¹⁶. There is also growing appreciation of the importance of post-ECMO functional outcome assessments.

Studies of long-term outcomes of ECMO survivors are limited by the heterogeneity in pathologies, outcome measures used, and age at follow-up ³⁰. Professional society and research networks are developing working groups aiming to collaborate in drafting recommendations for the use of standardized outcomes measures that are easy to implement and validated in several languages. Timing of follow-up would also need to be harmonized with age-specific recommendations by groups focusing on specific pathologies (e.g., congenital heart disease)¹¹⁷. Development of a core set of outcome measures to be used in future ECMO studies would aid in further understanding the risk factors and prognostic markers for unfavorable outcomes, and in identifying opportunities for intervention to improve patient outcomes ³⁰.

Recent advances in ECMO technology have attempted to minimize the pro-inflammatory and pro-thrombotic effects of exposure to the ECMO circuit through development of hollow fiber oxygenators, biocompatible coating for tubing and oxygenators, reduced surface area of ECMO circuits, and application of high blood flow rates that minimize blood stasis within the circuit ¹¹⁸. Emerging circuit surface modifications have the potential to decrease the risk of thrombus formation and thus of embolization, and to safely allow for reduced or complete elimination of systemic anticoagulation eventually, thus theoretically decreasing the risk for both ischemic embolic stroke and intracranial hemorrhage ¹¹⁹. Surface modifications include biomimetic surfaces (e.g., heparin, nitric oxide, direct thrombin inhibitors), biopassive surfaces (e.g., phosphorylcholine, albumin, etc) and endothelialization of blood contacting surfaces ¹¹⁹.

In summary, the burden of acute neurologic injury surrounding neonatal and pediatric ECMO support remains high, as do the rates of mortality and of unfavorable outcomes in survivors. Improving current neurologic monitoring methods and developing novel, real-time monitoring modalities during the high-risk periods leading to ECMO and during ECMO support are paramount for early detection of acute neurologic injury. Standardizing neurologic outcome assessments among survivors is an important next step for evaluating the true success of ECMO monitoring, management and long-term follow up. This in turn would promote early neuroprotective interventions with the ultimate goal to improve neurodevelopmental outcomes in this vulnerable population.