Clinical Outcomes of Extracorporeal Membrane Oxygenation Use in Patients With Intracranial Hemorrhage

Abstract

Background

Patients undergoing extracorporeal membrane oxygenation are at a high risk of developing intracranial hemorrhage as a neurological complication. Consequently, many physicians consider a history of intracranial hemorrhage as a relative contraindication for extracorporeal membrane oxygenation and are hesitant to use it in these patients, even in cases of acute severe heart or lung failure. This study aimed to examine the clinical outcomes of extracorporeal membrane oxygenation use in patients with intracranial hemorrhage.

Methods

We retrospectively obtained the medical records of patients diagnosed with intracranial hemorrhage who received extracorporeal membrane oxygenation owing to acute cardiopulmonary failure between January 2011 and July 2020. Data pertaining to patients’ characteristics and clinical outcomes were collected. Disseminated intravascular coagulation score and extracorporeal membrane oxygenation score before and after application of extracorporeal membrane oxygenation were also examined to observe trends.

Results

Eighteen patients were included. Ten had traumatic intracranial hemorrhage, and the most common indication for extracorporeal membrane oxygenation was acute respiratory distress syndrome. The 30-day survival rate was 72% (13 patients), and 61% (11 patients) survived to discharge. Two patients underwent neurosurgery due to worsening of intracranial hemorrhage. However, both were discharged without neurological deterioration.

Conclusion

A 72% survival rate was observed in extracorporeal membrane oxygenation patients with intracranial hemorrhage, suggesting that extracorporeal membrane oxygenation could be a viable option in patients with intracranial hemorrhage unresponsive to conventional therapy.

Graphical Abstract

INTRODUCTION

Extracorporeal membrane oxygenation (ECMO) evolved from the cardiopulmonary bypass used in heart surgery and has a simpler form comprising a membrane oxygenator, circuit, cannula, and blood pump. ECMO is used for patients with severe heart or pulmonary failure, whose survival rate is < 50%, with conventional treatments such as inotropic drug use and mechanical ventilation.1 Although ECMO use is increasing in cases of reversible severe cardiopulmonary failure, it carries a high risk of complications that could lead the patient to death. Renal dysfunction and ischemia of cannulated limbs are common side effects and are managed using continuous renal replacement therapy (CRRT) and distal perfusion of the affected limb. Among complications, intracranial hemorrhage (ICH) is the most devastating to physicians. The use of anticoagulants is essential to prevent thrombotic complications in ECMO; however, this increases bleeding tendency and ICH risk. ICH incidence in patients using ECMO varies across studies. Lorusso et al.2,3 conducted a retrospective analysis of the Extracorporeal Life Support Organization (ELSO)’s data registry, reporting ICH rates of 1.8% in venoarterial (V-A) ECMO2 and 3.6% in veno-venous (V-V) ECMO3 in adult patients. In another study examining the association between ECMO and brain injury using magnetic resonance imaging, neuroradiologic lesions were found in 52% of ECMO patients, with 7% having ICH, a much higher rate than that from the ELSO data.4 Several studies have reported low survival rates when ICH occurs in ECMO patients, ranging from 11% to 26%.2,3,5,6 Consequently, many physicians consider a history of ICH as a relative contraindication for ECMO and are reluctant to use ECMO in these patients, even in cases of acute severe heart or lung failure. Although many studies have examined the occurrence, risk factors, prognosis, and management of ICH as a complication of ECMO, few have evaluated ECMO outcomes in patients with a prior history of ICH. This study aimed to examine the clinical outcomes of ECMO use in patients previously diagnosed with ICH.

METHODS

Patients and variables

This study included adult patients diagnosed with ICH before ECMO insertion between January 2011 and July 2020. Cases in which ICH developed as a complication after ECMO treatment were excluded. All medical records, including clinical notes, laboratory results, and brain imaging data, were retrospectively collected.

The following data were collected: sex, age, body mass index, ECMO indication, ECMO duration (day), ECMO type, comorbidity, Charlson Comorbidity Index, and disseminated intravascular coagulation (DIC) and ICH scores.

The DIC score is used as a criterion for determining overt DIC and the degree of coagulopathy. In this study, the DIC score was measured based on the International Society on Thrombosis and Hemostasis scoring system,7 which includes platelet counts, prothrombin time, and fibrinogen and D-dimer levels.

The ICH score is a simple clinical grading scale used to predict mortality rate by scoring based on five factors: Glasgow Coma Score, age, ICH volume, intraventricular hemorrhage, and infratentorial origin of hemorrhage.8

ECMO insertion and anticoagulation strategy

The indication for ECMO insertion, ECMO type (V-V, V-A, venoarteriovenous [V-AV]), and initial anticoagulant management followed the standard protocol of our ECMO center, which is similar to the protocol used for patients without ICH. Cannulation of all ECMO patients was performed using the Seldinger technique in the cardiac catheterization or hybrid room in the emergency department with real-time fluoroscopy guidance through both femoral vessels.

The criteria for selecting the ECMO type are shown in Fig. 1. V-A ECMO was performed when the patient underwent cardiopulmonary resuscitation (CPR) or needed a norepinephrine concentration of ≥ 0.5 µg/min/kg and displayed an ejection fraction of ≤ 20%. ECMO use owing to pulmonary failure was performed when the PaO₂/FiO₂ ratio was < 100 at 100% oxygen concentration in mechanical ventilation or if blood pH was < 7.2 owing to CO₂ retention. If heart failure was not accompanied or blood pressure was maintained with vasopressors, V-V ECMO was performed. Otherwise, V-AV or V-A ECMO was performed depending on heart failure severity.

Fig. 1. Criteria for extracorporeal membrane oxygenation type selection in pulmonary failure.

P/F = PaO₂/FiO₂, Nepi = norepinephrine, CPR = cardiopulmonary resuscitation, EF = ejection fraction, VV = veno-venous, V-AV = venoarteriovenous, VA = venoarterial.

For initial loading anticoagulation, 2,000−5,000 units of heparin were injected with bolus right before cannulation. Nafamostat mesilate was used for continuous anticoagulant injection. Activated partial thromboplastin time was maintained between 60 and 80 seconds, using an infusion rate of 0.4–1.5 mg/kg/h of nafamostat mesilate. For patients requiring surgery during ECMO treatment, the continuous anticoagulant infusion was stopped on the day of surgery and restarted the next day.9

Patient monitoring and management during ECMO

All patients using ECMO were managed in the intensive care unit with a ventilator, and tracheostomy was routinely performed. Sedative drugs were administered following initial neurologic examination if no abnormalities were found to minimize patients’ irritable movements. Afterward, the brain condition was assessed by continuous monitoring of cerebral oximetry, pupil size, and reflex. A brain computed tomography was performed when neurological problems such as convulsions, pupil abnormalities, and brain-stem reflex loss or rapid decrease of the cerebral oximetry score without perfusion problems were observed. Laboratory monitoring cycle and transfusion or supplement standards are presented in Table 1. For antithrombin III, an initial loading dose of 2,000 IU was administered, followed by 500 IU every 6 hours for 3 days as a maintenance dose. CRRT was performed if a patient showed progressive oliguria (urine output < 0.5 cc/kg/h for > 6 hours).

Table 1. Patient laboratory monitoring and transfusion management criteria.

Laboratory monitoring		Maintain
Platelet	Every 8 hr	> 50,000–80,000	Platelet concentrates
INR	Every 24 hr	< 2.0	Fresh frozen plasma
Fibrinogen	Every 24 hr	> 100 mg/dL	Cryoprecipitate
CBC	Every 8 hr	Hct > 24–30%	Packed RBC
Antithrombin	Every 24 hr	> 70%	AT3

INR = international normalized ratio, CBC = complete blood cell count, Hct = hematocrit, RBC = red blood cell, AT3 = antithrombin III.

Ethics statement

The present study protocol was reviewed and approved by the Institutional Review Board (IRB) of Hallym University Sacred Heart Hospital (IRB No. 2020-09-017), which waived the need for informed consent due to the retrospective study deign.

RESULTS

Baseline characteristics of participants

This study included 18 patients, comprising seven male and 11 female individuals, with a mean age of 48 (range 19–74) years and a mean body mass index of 22.73 kg/m². The most common indication for ECMO was acute respiratory distress syndrome (ARDS) (12 patients), followed by stress cardiomyopathy (5 patients) and septic shock (3 patients). Nine patients received V-V ECMO, eight patients received VA, and one patient received V-AV. Ten patients underwent brain surgery owing to ICH, of whom eight underwent surgery before ECMO, and two underwent surgery during ECMO. A total of 13 patients underwent CRRT. The 30-day survival rate was 72% (13 patients), and the survival discharge rate was 61% (11 patients, Table 2).

Table 2. Analysis of characteristics between traumatic and spontaneous ICH.

Patients’ profiles		All (N = 18)	t-ICH (n = 10)	s-ICH (n = 8)
Sex, male		7 (39)	6 (60)	1 (13)
Age, yr		48 (19–74)	44 (19–74)	52 (28–74)
BMI, kg/m2		22.73	22.08	23.53
ICH sub-type
	EDH	1	1	0
	SDH	6	4	2
	SAH	12	7	5
	IVH	6	3	3
	IPH	9	4	5
	CCI (mean)	1.39	1.30	1.50
Brain surgery
	No	8	7	1
	Before-ECMO	8	2	6
	With ECMO	2	1	1
ECMO indication
	ARDS d/t lung contusion	5	5	0
	ARDS d/t pneumonia	5	2	3
	SCMP	3	1	2
	Septic shock	3	2	1
	ARDS with SCMP	2	0	2
ECMO mode
	V-V	9	6	3
	V-A	8	3	5
	V-AV	1	1	0
ECMO support time, day		11 (2–50)	9 (2–14)	13 (6–50)
CRRT
	Apply	13	7	6
	No	5	3	2
Survival
	Mortality	5 (28)	1 (10)	4 (50)
	30-day survival	2	2	0
	Survival to discharge	11 (61)	7 (70)	4 (50)

Values are presented as number (%) or mean (range).

ICH = intracranial hemorrhage, t-ICH = traumatic intracranial hemorrhage, s-ICH = spontaneous intracranial hemorrhage, BMI = body mass index, EDH = epidural hemorrhage, SDH = subdural hemorrhage, SAH = subarachnoid hemorrhage, IVH = intraventricular hemorrhage, IPH = intraparenchymal hemorrhage, CCI = Charlson Comorbidity Index, ECMO = extracorporeal membrane oxygenation, ARDS = acute respiratory distress syndrome, SCMP = stress induced cardiomyopathy, V-V = veno-venous, V-A = venoarterial, V-AV = venoarteriovenous, CRRT = continuous renal replacement therapy.

Traumatic and spontaneous ICH

Depending on the cause of ICH, several studies have explored various mechanisms of brain injury regarding cellular responses, heme toxicity, coagulation disorders, and vasospasm.10,11,12,13 Along with these mechanisms and characteristics of brain injury, traumatic ICH has a slightly better prognosis than spontaneous ICH.14 Therefore, patients were classified into two groups: traumatic and spontaneous.

Of the 18 patients, 10 had traumatic ICH, and 8 had spontaneous ICH. The traumatic ICH group had a high proportion of male individuals (60% vs. 13%) and lower mean age than the spontaneous ICH group (44 vs. 52 years). Brain surgery was performed in 3 of 10 patients with traumatic ICH, and 7 of 8 with spontaneous ICH, indicating a higher surgical intervention rate in the spontaneous ICH group. The mean ECMO support time was 9 and 13 days for the traumatic and spontaneous ICH groups, respectively, although one patient in the spontaneous ICH group died after 50 days of ECMO treatment. Excluding this case, the average duration was 8 days in the spontaneous ICH group. In the traumatic ICH group, the 30-day survival rate was 90%, and the survival-discharge rate was 70%, whereas in the spontaneous ICH group, both the 30-day survival and survival-discharge rates were 50% (Table 2).

DIC and ICH scores

The DIC score was measured from after the day of ECMO insertion to day 5, and the trend of the change was examined using a box-and-whisker plot. Median DIC scores remained stable at 5, 4, 4, 4, and 4 over this period, showing no change in score. The ICH score was evaluated at three points: upon ICH diagnosis, before ECMO, and after ECMO. The median ICH scores were 1.5, 2.5, and 2 at admission, before ECMO, and after ECMO, respectively (Fig. 2).

Fig. 2. Box-and-whisker plot of disseminated intravascular coagulation and intracranial hemorrhage score.

DIC = disseminated intravascular coagulation, ECMO = extracorporeal membrane oxygenation, ICH = intracranial hemorrhage.

Incidence of mortality cases

Seven patients died in the hospital without being discharged. Two of them survived > 30 days post-ECMO removal, and the causes of death were gastrointestinal bleeding and ascending cholangitis, respectively. Five people died within 30 days post-ECMO administration, of whom three died from worsening of sepsis. One patient underwent extracorporeal CPR but died the next day due to severe brain damage. The other also had severe brain damage; however, the patient’s guardians strongly requested ECMO treatment and waited for signs of consciousness for 7 days, although the patient died without recovery. To identify factors related to prognosis, we divided the participants into survivor and non-survivor groups for comparison, however, no significant factors were identified (Table 3).

Table 3. Comparison of characteristics in survivors and non-survivors.

Pre-ECMO variables, mean		Survivors (n = 11)	Non-survivors (n = 7)	P value
Sex, male		5 (45.4)	2 (28.5)	0.637
Age, yr		41 (19–74)	57 (39–74)	0.059
ICH type				0.335
	Spontaneous	4	5
	Traumatic	7	2
Midline shifting		1 (9)	2 (28.5)	0.528
ICH score		2.09	2.42	0.527
Platelet count, cells/uL		158.45	119.42	0.408
INR		1.35	1.37	0.828
Activated PTT, sec		49.64	60.04	0.262
Estimated GFR, mL/min/1.73 m2		60.85	54	0.668

Values are presented as number (%) or mean (range).

ECMO = extracorporeal membrane oxygenation, ICH = intracranial hemorrhage, INR = international normalized ratio, PTT = partial thromboplastin time, GFR = glomerular filtration rate.

DISCUSSION

Numerous studies on brain injury and ICH as complications of ECMO treatment have shown consistent results. When ICH occurs during ECMO, the prognosis is very poor. However, these results do not exhibit evidence that ECMO treatment should not be performed in patients with a history of ICH. Although sample sizes in ECMO studies are often small, our study suggests potential role of ECMO in both traumatic and spontaneous controlled ICH.

In the field of trauma and intensive care, ECMO use for traumatic lung injury is increasing, with reports of successful treatment even in cases involving traumatic brain injury.15 In this study, patients with traumatic ICH demonstrated a 30-day survival rate of 90% and survival discharge rate of 70%. Our results will further strengthen the claim that patients with trauma should be actively treated with ECMO.

No studies have specifically examined the use of ECMO in patients with spontaneous ICH. However, approximately 27% of patients with spontaneous ICH receiving mechanical ventilation develop ARDS,16 with a high mortality rate. Most critically ill patients with spontaneous ICH are admitted to the neurosurgery department, and in some of these patients, ARDS progresses to pneumonia or neurogenic pulmonary edema. In addition, since the neurosurgery department does not perform ECMO treatment, these patients are often overlooked for ECMO intervention. In this study, the survival rate of the patients in the spontaneous ICH group was 50% (four patients), which was much lower than that of those in the traumatic ICH group; however, many severely ill patients in the spontaneous ICH group underwent brain surgery, with death primarily caused by worsening sepsis due to failure to control infection rather than neurological complications. Therefore, there is insufficient evidence to classify spontaneous ICH as a contraindication for ECMO.

The high survival and low complication rates observed in this study may be attributed to the stability of the DIC score, an indicator of coagulopathy, following ECMO initiation. As presented in Table 1, our center actively addressed laboratory deficiencies such as cryoprecipitate and antithrombin III. We have shown in previous study that nafamostat mesilate could be considered as an alternative anticoagulant to heparin during ECMO.9 So, while its effectiveness remains debated, the use of nafamostat mesilate instead of heparin as an anticoagulant might have helped.17

Our findings suggest that conventional ICH scores did not worsen when anticoagulation was properly administered and treated to prevent thrombotic complications. Collaboration with neurosurgery teams is essential to actively consider the application of ECMO in patients with ICH who are unlikely to experience further bleeding, potentially through coil embolization or surgical methods.

This study has several important limitations. First, it is a single-center, retrospective study, which inherently limits the generalizability of the findings. The small sample size also poses limitations in ensuring statistical significance, and the findings should be interpreted with caution. Larger, multicenter studies are necessary to validate these results in more diverse patient populations. Owing to the fact that patients using EMCO can present with varying degrees of severity, the clinical condition of these patients can differ, indicating that there will be many patient groups who need ECMO in the neurosurgery department, and more patients can be saved through active inter-department consultation. Further multicenter studies are warranted to explore this issue comprehensively. Second, there is limited information regarding the long-term safety and side effects of nafamostat mesilate. Some studies highlighted some potential adverse effects, such as hyperkalemia, anaphylaxis, and other coagulation-related concerns,18 and the long-term impact of nafamostat mesilate remains unclear. Further research is needed to evaluate its safety profile over extended periods. Third, this study was conducted without a control group, which makes it difficult to draw definitive conclusions regarding the relative effectiveness of nafamostat mesilate compared to heparin. Future randomized controlled trials large-scale multicenter studies are required to directly compare the safety and efficacy of nafamostat mesilate with heparin in ECMO patients, particularly those diagnosed with ICH.

In conclusion, the potential for ICH should always be considered in patients undergoing ECMO treatment, and appropriate management strategies should be in place. However, not considering this may lead to dismissal or delay of ECMO procedures for patients with a history of ICH, thereby depriving them of a chance to survive. Despite the small sample size of exploratory study, we showed high survival rates (72%) and low neurologic complications rate (17%), with only one complication-related mortality case. This study may inspire other ECMO centers to expand the potential role of ECMO in patients with ICH, facilitating further research in this area.