Brain Injury Plasma Biomarkers in Patients on V-A ECMO: A Pilot Prospective Observational Study

Abstract

Introduction:

Early diagnosis of acute brain injury (ABI) is critical for patients on veno-arterial extracorporeal membrane oxygenation (V-A ECMO) to guide anticoagulation strategy; however, neurological assessment in ECMO is often limited by patient sedation.

Methods:

In this pilot study of adults from June 2018 to May 2019, plasma samples of glial fibrillary acidic protein (GFAP), neurofilament light chain (NFL), and Tau were collected daily after V-A ECMO cannulation and measured using a multiplex platform. Primary outcomes were occurrence of ABI, assessed clinically, and neurologic outcome, assessed by modified Rankin Scale (mRS).

Results:

Of 20 consented patients (median age=48.5 years; 55% female), 8 (40%) had ABI and 15 (75%) had unfavorable neurologic outcome at discharge. 10 (50%) patients were centrally cannulated. Median duration on ECMO was 4.5 days (IQR: 2.5–9.5). Peak GFAP, NFL, and Tau levels were higher in patients with ABI vs. without (AUC = 0.77; 0.85; 0.57, respectively) and in patients with unfavorable vs. favorable neurologic outcomes (AUC = 0.64; 0.59; 0.73, respectively). GFAP elevated first, NFL elevated to the highest degree, and Tau showed limited change regardless of ABI.

Conclusion:

Further studies are warranted to determine how plasma biomarkers may facilitate early detection of ABIs in V-A ECMO to assist timely clinical decision-making.

1. Introduction

Extracorporeal membrane oxygenation (ECMO) is an increasingly utilized intervention that provides emergency respiratory and circulatory support to patients with refractory pulmonary and cardiac failure¹. Despite the demonstrated survival benefits of ECMO compared to standard care, the survival rate among ECMO patients remains suboptimal, with a reported 58% survival rate to discharge², across both veno-arterial ECMO (V-A ECMO) and veno-venous ECMO (V-V ECMO) patients. In studies specifically looking at patients on V-A ECMO support, one of the leading causes of morbidity and mortality is acute brain injury (ABI), which has been reported to occur in up to 20% of adult V-A ECMO patients and confers a two-fold increase in mortality risk^3,4. Common neurological complications of V-A ECMO include intracranial hemorrhage, ischemic stroke, seizures, hypoxic-ischemic brain injury, and cerebral edema⁵. However, given the abundance of patients on heavy sedation with unreliable neurological examination, early diagnosis of ABI is challenging. As V-A ECMO patients are often too unstable to transport for neuroimaging studies out of the intensive care unit (ICU), there is a clinical need for easy-to-use, bedside-based tools in the ECMO population. Thus, the purpose of blood-based biomarkers is not to replace, but rather to predict ABIs that encourage earlier performance of neuroimaging⁶. Additionally, while studies have explored these biomarkers for children on V-A ECMO, there is little data exploring the utility of these biomarkers for adult patients on V-A ECMO⁷.

Glial fibrillary acidic protein (GFAP), neurofilament light chain (NFL), and Tau are three sensitive and specific markers for neuronal injury. Their early elevation predicted unfavorable neurological outcomes in patients with cardiac arrest as well as neurodegenerative diseases^8,9. Glial fibrillary acidic protein (GFAP) is a cytoskeletal protein that is specifically expressed in astrocytes¹⁰. Elevated levels of GFAP in the blood have been shown to be a marker of astroglial activation and can be used to assess brain injury¹⁰. GFAP levels are higher in patients with intracranial hemorrhage than in patients with ischemic events, which have a longer duration before peak GFAP is reached¹¹. This suggests that GFAP may be a useful marker for identifying acute brain injuries. Neurofilaments confer structural stability to neurons and are present in dendrites, neuronal soma, and axons¹². Elevated levels of NFL in blood and cerebrospinal fluid indicate axonal injury^12,13. Similarly, NFL levels were significantly higher in unfavorable neurological outcome (vs. favorable) after hypoxic-ischemic brain injury in OHCA, especially at 1–3 days with high sensitivity and specificity for 6-month neurological outcomes^8,9. Serum tau stabilizes microtubules and is found in neurons¹⁴. Tau protein at 72 hours was accurate in predicting 6-month neurological outcome with the area under the receiver operating characteristic curve (AUROC) of 0.91 in out-of-hospital cardiac arrest (OHCA) patients^8,9. Despite the promising results of their accuracy in predicting ABIs and neurological outcomes in OHCA, these plasma biomarkers have not been tested in adult patients on ECMO support.

In this context, we hypothesized that specific blood-based biomarkers, namely GFAP, NFL, and Tau, could serve as early indicators of brain injury and thus inform early neuroimaging during V-A ECMO. Herein, we conducted a pilot prospective observational study on three brain-specific blood-based biomarkers (GFAP, NFL, and Tau) in adult patients with V-A ECMO support.

2. Materials and Methods

2.1. Study Design and Protocol

This is a preliminary report of our ongoing prospective observational cohort study on a series of ECMO patients admitted to a tertiary care academic hospital. Patients who required admission to the ICU for ECMO were prospectively enrolled. After the patient was placed on ECMO, there was a 24-hour window for the patient’s family to provide consent. We analyzed 20 patients who were consented within 24 hours and underwent plasma sampling once daily over the course of their ECMO duration. Patients were alerted to the research assistant who was available 24 hours a day and assisted in consenting patients, collecting the samples, and evaluating and tracking clinical data over time. These patients were followed until discharge. We have previously described our standard neuromonitoring protocol as delineated by Cho et. al and Ong et. al. which consists of baseline neurologic examinations including Glasgow Coma Scores (GCS), Acute Physiology and Chronic Health Evaluation II (APACHE II), and Sepsis-related Organ Failure Assessment (SOFA) scores, and protocols for serial imaging by transcranial doppler, electroencephalography, somatosensory evoked potentials and head computed tomography (CT) or magnetic resonance imaging (MRI), as clinically indicated^15,16. Clinical assessment following this standardized, comprehensive neuromonitoring protocol allowed for clinical diagnoses of specific types of ABI and appropriate patient treatment. Inclusion criteria included patients with V-A ECMO support aged ≥ 18 years. Exclusion criteria included pregnant patients and those aged ≤ 18 years. Patients on V-V ECMO were also excluded due to differing risk factors and mechanisms of ABI between V-A and V-V ECMO patients. Pre-existing atherosclerotic disease burden was indirectly assessed via past medical history, such as the presence of diabetes, hypertension, prior ischemic stroke, and/or antiplatelet therapy before hospitalization. This study was approved by the Johns Hopkins institutional review board.

2.2. Sample Collection, Processing, and Storage Protocol

Standard clinical care for patients placed on ECMO was followed, including physical examination at bedside on the first day of ECMO and close follow-up until hospital discharge or death. Venous blood samples (5 mL in sodium citrate 3.2%) were ideally collected once within 24 hours after initiation of ECMO, and then once daily thereafter. If the patient’s family consented within the 24 hour window after ECMO cannulation but a sample had not been taken within 24 hours, these participants were still included and time of first plasma collection was noted. Samples were spaced to every other day after the first 7 days with ECMO support. Daily samples were scheduled at the same time as prior sample testing to ensure consistency. Across different participants, samples were not taken at the same time every day due to different ECMO cannulation start-times. All samples were centrifuged at 3000 RPM for 8 minutes and divided into four aliquots, with the first aliquot containing 250 μL, the second aliquot containing 500 μL, and the remainder evenly split into the other two aliquots. These tubes were labeled based on patient number, sample number, and aliquot number – along with the date that the samples were collected. Plasma was extracted from the samples, taking care not to extract red blood cells. The exact amounts extracted were recorded, and each tube was again labeled based on patient number, sample number, and aliquot number. These samples were stored in a −80 °C freezer. GFAP, NFL and Tau were assayed using a commercial multiplex assay (Meso Scale Discovery, Gaithersburg, MD) using the assay protocol and quantified using a SECTOR Imager 6000 plate reader (Meso Scale Discovery). The upper limit of detection for GFAP, NFL, and Tau were 25,000, 7,000, and 300, respectively. The lower limit of detection for GFAP, NFL and Tau were 1.14, 0.563, 0.014, respectively. The average coefficient of variation for GFAP, NFL, and Tau were 7%, 4%, and 16%, respectively. All three analytes were quantifiable in each sample.

2.3. Definitions and Outcomes

A comprehensive collection of pre-ECMO patient history, including demographics, past medical history, ECMO and laboratory variables (creatine, platelet count, lactate, and transaminase values) were recorded for each patient. Drawing from our definitions used in a prior study of ABIs in patients with ECMO support, ABIs in this study were defined as having experienced subdural hemorrhage, subarachnoid hemorrhage, intracerebral hemorrhage, ischemic stroke, hypoxic-ischemic brain injury, and brain death diagnosed by relevant neuroimaging protocols (head CT and MRI)¹⁶. While clinical assessment and neurologic examination could inform the use of neuroimaging, only neuroimaging itself could make a diagnosis of ABI based on the definitions provided previously. While CT is recommended for all ECMO patients, often they are too unstable to transport. Additionally, some patients (such as postcardiotomy shock patients) may not present with neurological issues on initial exam and therefore do not warrant neuroimaging. Therefore, while recommended, it was not standard practice for all patients to have neuroimaging prior to or right after cannulation. We utilized our institution’s existing prospectively collected ECMO database for baseline and in-hospital data^15,16. Patient charts were manually reviewed and validated for the timing of an ABI following ECMO cannulation. Favorable neurological outcome was defined as modified Rankin Scale (mRS) ≤ 3 and unfavorable neurological outcome as mRS ≥ 4 at discharge.

2.4. Statistical Analysis

Baseline patient characteristics, patient demographics, and neurologic outcomes were compared between patients with ABI vs. without using Chi-squared tests and Fischer’s exact texts for categorical variables. Mann-Whitney U test was used for continuous variables. Statistical significance level was set at p<0.05, and analyses were performed using Stata 17 (College Station, TX, USA). Serum biomarker peaks were plotted for each patient during ECMO support. Boxplots were created to visualize the differences in peak biomarker concentrations between patients with and without ABI. Sensitivity and specificity analyses were conducted. C-statistics, also known as the area under the receiver operating characteristic (ROC) curve, were computed for ABI and unfavorable neurologic outcome. These curves were pictographically demonstrated using GraphPad Prism version 9.0 (San Diego, CA, USA). To examine temporal changes in each biomarker concentration and the ABI timing, peak biomarker values were plotted for each day following ECMO cannulation for six out of eight ABI patients (Figure 3). For the time-series analysis, two patients were excluded due to rapid patient mortality after cannulation, yielding no insight into biomarker fluctuation over time. Time-series analyses were performed to characterize the time of diagnosis of ABI in relation to biomarker levels, and the results were displayed using Microsoft Excel 2019 (Redmond, WA, USA).

Figure 3.

Time-series plots of peak biomarker levels for six different patients over the course of ECMO cannulation. Red line indicates diagnosis of an ABI. 3A) Patient with ABI on Day 3. 3B) Patient with an ABI on Day 7 and Day 9. 3C) Patient with unidentifiable ABI. 3D) Patient with ABI on Day 7. 3E) Patient with ABI on Day 2. 3F) Patient with ABI on Day 4.

3. Results

Our pilot study included a patient cohort consisting of 20 patients (median age=48.5 years, 41.5–62.0), with 55% (n=11) being female. Data on number of patients who were screened or refused to participate was not collected. Median number of days from symptom onset to ECMO cannulation was 3 days (IQR: 1–12.25). All 20 patients were supported with V-A ECMO) with 10 patients (50%) with central cannulation (50% peripheral V-A ECMO). The primary indications for ECMO were medical cardiogenic shock (60%, n=12), post-cardiotomy shock (35%, n=7), and ECPR (5%, n=1). The median duration on ECMO was 4.5 days (IQR: 2.5–9.5). All patients had a heparin gtt, and while no patients had a previous diagnosis of coagulopathy, bleeding was a common occurrence, with the mean number of packed red blood cell transfusions received being 17.6 and the mean number of platelet transfusions received being 3.2. Of the 20 patients, 15 patients (75%) had a head CT performed. ABI occurred in 40% (n=8) of the patients during ECMO support. All patients with an ABI were of the 15 who had brain imaging completed. As specified by diagnoses made in the electronic medical record (EMR), there were 4 instances of ischemic stroke, 1 instance of hypoxic-ischemic brain injury, 1 instance of subdural hematoma, and 1 instance of brain death. In one patient, specific day of ABI (patient C) could not be identified. One patient had an ABI (occurrence of ICH, SAH, and SDH) prior to ECMO cannulation, and therefore was not included in the ABI group. In total, 12 patients had care withdrawn due to family decision (n=10), non-CNS hemorrhage (n=1), and primary diagnosis being incompatible with life (n=3). The remaining patient characteristics are displayed in Table 1, stratified by the presence of ABI.

Table 1.

Demographics, past medical history, pre-ECMO variables, and ECMO variables of veno-arterial (V-A) extracorporeal membrane oxygenation (ECMO).

	Patients without ABI (n=12)	Patients with ABI (n=8)	p-value
Demographics
Age, years	49 (43–68)	48 (42–58)	0.67
Male	6	3	0.67
Ethnicity			0.33
White	7	4
Black	2	4
Hispanic	2	0
Asian	1	0
Medical History
Prior Ischemic Stroke	2	0	0.49
ICH History	0	0
Hypertension	8	7	0.60
Diabetes Mellitus	3	0	0.24
Congestive Heart Failure	3	5	0.094
Chronic Kidney Disease	2	1	1
Atrial Fibrillation	5	3	1
Antiplatelet therapy before hospitalization	7	4	1
Anticoagulant therapy before hospitalization	4	2	1
Pre-ECMO Variables
Ionotropic or Vasoactive Support	8	5	1
GCS	15 (11–15)	15 (9–15)	0.84
pH	7.24 (7.00–7.34)	7.26 (7.20, 7.40)	0.53
pCO2, mmHG	43 (36–61)	43 (39–52)	0.96
pO2, mmHG	110 (44–265)	261 (175–396)	0.062
Bicarbonate, mmol/L	18 (16–22)	20 (18–21)	0.21
O2 Saturation	87 (64–100)	99 (98–99)	0.12
Primary ECMO Indication
Cardiogenic Shock	8	4	0.65
Post-cardiotomy Shock	4	3	1.00
ECPR	0	1	0.40
ECMO Variables (Day 1)
Serum Creatinine, mg/dL	1.35 (1–1.65)	1.65 (1.1–2)	0.33
Platelets, × 10^9/L	63 (49–93)	87 (50–206)	0.49
Lactate, mmol/L	4 (2–7)	4 (3–6)	0.29
AST, U/L	468 (93–1028)	121 (109–163)	0.11
ALT, U/L	98 (26–474)	23 (11–111)	0.07
APACHE II Score	30 (22–35)	29 (25–32)	1.00
SOFA Score	12 (10–15)	11 (10–12)	0.53
Neuroimaging Received			0.16
CT	8	7
MRI	0	2
ABI Type
Ischemic stroke	-	6	-
HIBI	-	1	-
Subdural hematoma	-	1	-
Brain death	-	1	-
ECMO Duration, days	4 (3–8)	8 (3–14)	0.24
ER presentation to time of first blood draw, days	3 (1–10.75)	4 (0.75 – 17.25)	0.87
Time of ECMO cannulation to first sample drawn			0.48
0–8 hours	1	0
9–16 hours	2	0
17–24 hours	6	6
>24 hours	3	2

ABI = acute brain injury, ICH = intracerebral hemorrhage, GCS = Glasgow Coma Score, ECPR = Extracorporeal cardiopulmonary resuscitation, AST = aspartate aminotransferase, ALT = alanine aminotransferase, APACHE II = Acute Physiology and Chronic Health Evaluation II, SOFA = Sequential Organ Failure Assessment, CT = computerized tomography, MRI = magnetic resonance imaging, HIBI = hypoxic-ischemic brain injury, ER = emergency room. Results presented as N for categorical measures and median (IQR) for continuous variables.

3.1. Brain Injury Biomarkers and ABI

The median peak concentration values and their interquartile ranges for GFAP, NFL, and Tau were 382.3 pg/ml (100.8–5,516.4), 8,306.7 pg/ml (2,338.9–13,577.2), and 1,278.4 pg/ml (318.7–3,447.4), respectively. The median time to peak concentration was 3 days (IQR: 2.0–6.3), 7 days (IQR: 2.0–10.0), and 2 days (IQR: 1.0–6.8), respectively. Median peak GFAP and NFL concentrations were significantly higher for patients who developed ABI vs. without (5167 pg/ml vs. 160.9 pg/ml, p=0.047; 13577 pg/ml vs. 3424 pg/ml, p=0.007), respectively. Although the difference in median peak Tau concentrations did not reach statistical significance, a trend favored higher median peak Tau concentrations in patients with ABI vs. without (1871 pg/ml vs. 439.7 pg/ml, p=0.62) (Figure 1). The C-statistics calculated by plotting sensitivity and specificity for each biomarker yielded AUCs of 0.77, 0.85, and 0.57 for GFAP, NFL, and Tau, respectively (Figure 2).

Figure 1.

125 samples collected for each biomarker. 1A) Peak GFAP concentration for patients with and without ABI. 1B) Peak NFL concentration for patients with and without ABI. 1C) Peak Tau concentration for patients with and without ABI. 1D) Peak GFAP concentration for patients with and without favorable neurologic outcome. 1E) Peak NFL concentration for patients with and without favorable neurologic outcome. 1F) Peak Tau concentration for patients with and without favorable neurologic outcome. Unfavorable neurologic outcome status as assessed by mRS score>3. * denotes p < 0.05. ** denotes p <0.01. ns denotes not significant.

Figure 2.

Sensitivity and specificity analysis displayed as Receiver Operating Characteristic (ROC) Curves for 20 patients for each biomarker. 2A) ROC curve for GFAP with respect to unfavorable neurologic outcome and ABI. 2B) ROC curve for NFL with respect to unfavorable neurologic outcome and ABI. 2C) ROC curve for Tau with respect to unfavorable neurologic outcome and ABI.

3.2. Brain Injury Biomarkers and Neurological Outcomes

The distribution of mRS is shown in Supplemental Figure 1. Overall, 15/20 patients (75%) had unfavorable neurological outcomes at discharge. The difference in median peak concentrations did not reach statistical significance for any biomarker with respect to unfavorable vs. unfavorable neurological outcome. However, trends did show higher biomarker levels in the unfavorable neurological outcome group vs. favorable group (GFAP: 462.6 pg/ml vs. 178.5 pg/ml, p=0.39; NFL: 9048 pg/ml vs. 6090 pg/ml, p=0.61; Tau: 1999 pg/ml vs. 327.8 pg/ml, p=0.14) also shown in Figure 1. The C-statistics calculated by plotting sensitivity and specificity for each biomarker yielded AUCs of 0.64, 0.59, and 0.73 for GFAP, NFL, and Tau, respectively (Figure 2).

3.3. Timing of Injury and Biomarker Peak

A total of 125 blood samples were collected over a median of 4.5 days (IQR: 2.5–9.5) per patient. Notable descriptive findings from graphing the temporal relationship of biomarkers with individual patients are as follows (Figure 3). For patient A (ischemic stroke), GFAP levels rose prior to the diagnosis of the injury, peaked on the day of diagnosis (Day 3), and continued to stay elevated thereafter. Patient B (seizure and SDH respectively) had 2 instances of ABI (Day 7 and 9) during ECMO, and GFAP and NFL levels began to rise immediately upon cannulation, with GFAP plateauing around the day of first ABI diagnosis and NFL continuing to rise throughout ECMO duration. Patient C (punctuate infarctions involving the right corona radiata and left parietal subcortical white matter) could not have a day of ABI diagnosis reliably ascertained and had no change in GFAP or Tau but around day 7 saw a rapid rise in NFL levels that grew exponentially until day 11. Patient D (acute brain infarcts in right cerebellum and left hippocampus) saw rising NFL levels starting from cannulation that saw a relative peak on day of diagnosis (Day 7) and continued to rise at a lower rate until Day 9. Patient E (ischemic stroke) saw a spike in GFAP levels upon cannulation that peaked on the day of diagnosis of ABI (Day 2) and the values of GFAP fell shortly thereafter and remained low until Day 16, while NFL levels started to rise immediately at cannulation and rapidly rose following injury on Day 2 until Day 16. Patient F (ischemic stroke) saw a relative peak of GFAP and NFL on day of ABI diagnosis (Day 4) which then decreased until around Day 6 and then began to rise again until Day 14, with NFL levels seeing the highest rise between the 3 biomarkers. The final two patients with an ABI (ischemic stroke and hypoxic-ischemic brain injury respectively) did not have temporal trends analyzed due to only two days of biomarker data. It is important to note that we do not know the exact timing of ABI in V-A ECMO patients due to heavy sedation and lack of neurological examination. Thus, this time series analysis replicated in larger cohorts may provide valuable insight in understanding the timing of ABI regardless of the radiographical test. Further research exploring the temporal relationship between biomarker changes is necessary to generate clinically useful associations.

Overall, GFAP typically elevated first, NFL elevated to the highest degree, and Tau showed limited change throughout the ECMO process regardless of ABI. The diagnosis of ABI co-occurred with a spike in either GFAP or NFL in five out of six cases analyzed. NFL levels increased starting from ECMO cannulation until day 5 even in the non-ABI group but grew at a faster rate between days 3 and 5 for patients that developed an ABI (Supplemental Figure 2). GFAP levels were on aggregate higher in the ABI group than in the non-ABI group and Tau showed no differences across cohorts (Supplemental Figure 2).

4. Discussion

ECMO is a life-saving cardiopulmonary intervention that carries risk for ABI such as hypoxic-ischemic brain injury, intracranial thromboembolic or hemorrhagic events, and brain death^3,4. Early diagnosis of ABI can guide clinicians on timely intervention, including cessation and judicious resumption of anticoagulation. Several anticoagulation strategies have been studied to reduce thromboembolic and hemorrhagic events, with mixed results^17,18. However, the means for assessing neurological status in V-A ECMO is limited due to the routine use of sedation and limited ability for patient transport. Currently, no brain injury biomarkers are routinely used in clinical practice despite ongoing concerted efforts^19–22. Blood biomarkers have a series of advantages including low costs, ease of access, quick processing times, and minimal technical expertise necessary for accurate and precise quantification.

The principal findings of this pilot study in adult V-A ECMO patients are that GFAP, NFL, and Tau are elevated in patients with an ABI vs. without. Similarly, the levels of these biomarkers were higher in patients with unfavorable neurological outcomes vs. favorable neurological outcome at discharge. This prospective pilot study builds upon previous research using these biomarkers in patients with cardiac arrest for their utility in neurological prognostication^8,9,19. To our knowledge, our study is the first to investigate these biomarkers in the context of adult V-A ECMO patients. Further research is necessary to build upon this pilot study to validate and determine these findings and to provide useful thresholds that can provide valuable insight regarding V-A ECMO patient status especially while they are sedated, and routine neurologic examination is limited.

4.1. GFAP

Median peak GFAP concentrations were significantly higher for patients who developed ABI vs. without. This corroborates prior research on GFAP outside of ECMO setting, which demonstrates correlation between GFAP levels and the severity of neurologic deficits in ischemic stroke patients²³. Upon analysis of GFAP levels, our data revealed a notable trend: for three patients out of the six that developed ABI, GFAP levels exhibited a relative peak on the same day as their brain injury diagnosis. Additionally, GFAP levels were markedly higher earlier (between days 1–5) of ABI patients than their non-ABI counterparts. This is consistent with prior findings with respect to TBI, as GFAP has repeatedly shown to rise 1 hour after injury and peak 20–24 hours after injury^24–26. This trend could be mechanistically explained by the elevation of GFAP in response to astrocyte-mediated repair processes following ABI¹⁰. These repair processes might have already been initiated prior to the clinical diagnosis of ABI, accounting for the earlier observed increase in GFAP levels leading up to a peak at the time of clinically diagnosed ABI. Given GFAPs biological role, as well as the findings of previous studies showing that GFAP is elevated following brain injury, it has potential for being a useful biomarker for brain injury occurrence and severity^27,28.

4.2. NFL

Median peak NFL concentrations were significantly higher for patients who developed ABI vs. without. Despite peak NFL levels occurring at ~7 days, NFL levels increased starting from ECMO cannulation until day 5 even in the non-ABI group and grew at a faster rate between days 3 and 5 for patients that developed an ABI (Figure A2). One of the major causes of ABI is right neck vessel cannulation that can causes disturbances to cerebral blood flow^29,30. Rising NFL levels for all patients, even those that did not develop an ABI, may be an indication of the severity of altered cerebral blood flow and may be explained by the release of NFL in the context of neuronal death and axonal damage¹³. However, patients with ABI saw a more rapid increase in NFL levels and saw a greater peak at the time of diagnosis, which could aid clinicians towards a quicker and more accurate diagnosis of ABI. These findings corroborate previous research on cardiac arrest patients which indicate that NFL levels are associated with hypoxic-ischemic brain injury in children and worsening outcomes in adults^8,31. Further research with sufficient sample size is warranted to investigate clinically useful cutoffs and better understand temporal associations between peak NFL levels and the severity of ABI, as well as its impact on neurological outcomes.

4.3. Tau

Although the difference in median peak Tau concentrations did not reach statistical significance, a trend favored higher median peak Tau concentrations in patients with ABI vs. without. While tau has been most heavily studied in its misfolded aggregate form in Alzheimer’s disease and other forms of dementia, tau protein release can potentially be used as a biomarker indicating axonal damage³². Tau has demonstrated potential significance for neuro-prognostication after cardiac arrest, but there is a paucity of data in ECPR and ECMO⁹. Our results indicate Tau protein having poorer sensitivity and specificity in association to ABI than GFAP and NFL. After acute brain injury, GFAP and NFL may be released into the bloodstream more rapidly compared to tau protein. This early release allows for prompt detection and measurement of these biomarkers, enabling their potential use in early prognostication. In contrast, tau protein release may occur later in the injury cascade, reducing its immediate predictive value.

GFAP, NFL, and Tau represent a small subset of a larger and broader set of biomarkers that may inform clinicians of patient’s neurological status and guide earlier neuroimaging or changes in anticoagulation strategy. With this study, we hope to demonstrate the feasibility and utility of studying biomarkers in adult V-A ECMO patients and urge continued research to provide a holistic understanding of the role of biomarkers in the context of broader clinical care.

5. Limitations

The are several limitations of our study. Like other pilot studies on ECMO, this study only has 20 patients and information on number of patients who were screened or refused to participate was not collected. Due to the preliminary nature of this study and the heterogeneous population with respect to the varying indications for ECMO, caution should be taken when interpreting the findings. Future research with larger cohorts is necessary to replicate these findings and better account for potential confounding factors, such as indication for ECMO. Additionally, with a sufficient sample size, a meaningful threshold analysis could be performed—especially as it relates to the predictive values for biomarkers to predict ABI and poor outcomes. The sample size limited our ability to control for multiple parameters and confounders. For example, it should be noted that biomarkers may elevate differently in isolated ischemia versus intracranial bleeding, and including them together does not allow for stratification with respect to type of ABI. This, amongst other factors, must be addressed in a larger, future study. Although our methodology was designed to mitigate confounding variables, we acknowledge that the variance of pre-existing atherosclerotic disease burden across the cohort may limit the generalizability of our findings. While we used past medical history as a proxy for pre-existing atherosclerotic disease burden, we acknowledge that this does not capture disease burden as well as direct visualization of the cranial vessels. Thus, future studies should both explore and factor in the impact that pre-existing atherosclerotic disease has on the results presented in this study. Secondly, we do not have control non-ECMO groups or blood samples prior to ECMO cannulation. As V-A ECMO is mostly an emergent cannulation, obtaining blood from these patients is not usually possible. Thirdly, although our center has a standardized neuromonitoring protocol that provides better ABI diagnosis ability, performing neuroimaging in a timely manner poses a challenge in ECMO patients with multiple vasoactive medications and often with open chest^15,16. For this, our institution currently is performing a study using a portable brain MRI at bedside (SAFE MRI ECMO study) for ECMO patients, which will improve our current study^33,34. Lastly, the mortality is high in V-A ECMO patients, and thus, mRS is skewed towards an unfavorable neurological outcome, limiting discrimination from favorable outcomes. Our ongoing efforts to conduct a multicenter study on this topic can address this limitation.

6. Conclusions

GFAP, NFL, and Tau represent promising novel biomarkers at bedside for predicting ABI and neurological outcomes in adult V-A ECMO patients where accurate neurological examination is limited. Our findings from this pilot study provide valuable insights into the dynamic and temporal changes of these biomarker levels in relation to ABI. Further research is crucial to refine the clinical applications of promising biomarkers, such as identifying signals for various types of brain injury and determining their utility and prognostic values in adult V-A ECMO patients.