Impact of Pulse Pressure on Acute Brain Injury in Venoarterial ECMO Patients with Cardiogenic Shock During the First 24 Hours of ECMO Cannulation: Analysis of the Extracorporeal Life Support Organization Registry

Background: Low pulse pressure (PP) in venoarterial-extracorporeal membrane oxygenation (VA-ECMO) is a marker of cardiac dysfunction and has been associated with acute brain injury (ABI) as continuous-flow centrifugal pump may lead to endothelial dysregulation. Methods: We retrospectively analyzed adults (≥18 years) on “peripheral” VA-ECMO support for cardiogenic shock in the Extracorporeal Life Support Organization Registry (1/2018–7/2023). Cubic splines were used to establish a threshold (PP≤10 mmHg at 24 hours of ECMO support) for “early low” PP. ABI included central nervous system (CNS) ischemia, intracranial hemorrhage, brain death, and seizures. Multivariable logistic regressions were performed to examine whether PP≤10 mmHg was associated with ABI. Covariates included age, sex, body mass index, pre-ECMO variables (temporary mechanical support, vasopressors, cardiac arrest), on-ECMO variables (pH, PaO2, PaCO2), and on-ECMO complications (hemolysis, arrhythmia, renal replacement therapy). Results: Of 9,807 peripheral VA-ECMO patients (median age=57.4 years, 67% male), 8,294 (85%) had PP>10 mmHg vs. 1,513 (15%) had PP≤10 mmHg. Patients with PP≤10 mmHg experienced ABI more frequently vs. PP>10 mmHg (15% vs. 11%, p<0.001). After adjustment, PP≤10 mmHg was independently associated with ABI (adjusted odds ratio [aOR]=1.25, 95% confidence interval [CI]=1.06–1.48, p=0.01). CNS ischemia and brain death were more common in patients with PP≤10 mmHg vs. PP>10 mmHg (8% vs. 6%, p=0.008; 3% vs. 1%, p<0.001). PP≤10 mmHg was associated with CNS ischemia (aOR=1.26, 95%CI=1.02–1.56, p=0.03) but not intracranial hemorrhage (aOR=1.14, 95%CI=0.85–1.54, p=0.38). Conclusions: Early low PP (≤10 mmHg) at 24 hours of ECMO support was associated with ABI, particularly CNS ischemia, in peripheral VA-ECMO patients.


Introduction
2][3][4] Acute brain injury (ABI), including intracranial hemorrhage (ICH), ischemic stroke, and hypoxic-ischemic brain injury (HIBI) occurs in up to 20% of adults on VA-ECMO support and is associated with increased mortality risk. 5Blood pressure variables, such as pulse pressure (PP), de ned as the difference between systolic (SBP) and diastolic (DBP) blood pressure, have been shown to be important surrogate markers of cardiovascular function in patients on mechanical circulatory support. 6 7, 8-ECMO operates with a continuous-ow centrifugal pump, which is associated with endothelial dysregulation/dysfunction 9,10 Although the precise mechanism is not entirely understood, this ensuing endothelial cell dysregulation resulting from nonpulsatile ow predisposes patients to neurological injury such as ABI.Therefore, PP may be a good surrogate marker for predicting neurological outcomes in ECMO patients.A recent study demonstrated PP < 20 mmHg within 12 hours of ECMO cannulation was associated with ABI in 123 VA-ECMO patients. 11Still, this study was limited by small sample size, from single center, and including central VA-ECMO and post-cardiotomy shock patients who are at higher predisposition of ABI. 12 Furthermore, peripheral VA-ECMO patients have hemodynamics states that are associated with vascular and perfusion abnormalities 13 and are thus an important population to investigate the association between PP and ABI.
Using the largest registry of ECMO patients globally, the Extracorporeal Life Support Organization (ELSO) Registry, we sought to investigate the association between early PP and ABI in peripheral VA-ECMO patients.We hypothesized that low PP in the rst 24 hours of ECMO support was independently associated with higher occurrence of ABI.

Study design and population
This study was approved by the Johns Hopkins Hospital Institutional Review Board with a waiver of informed consent since this was a retrospective observational study (IRB00216321).The ELSO Registry is an international multicenter registry from over 500 ECMO centers. 14The Registry collects demographics, pre-ECMO comorbidities, pre-ECMO and on-ECMO hemodynamic and arterial blood gas (ABG) information, on-ECMO neurological and other systemic complications, and outcomes such as mortality.15 Comorbidity information was recorded using the International Classi cation of Diseases, 10th Revision (ICD-10) codes.
We included patients who were 1) 18 years of age or older; and 2) supported with "peripheral" VA-ECMO and diagnosed with CS from 2018-2023.We excluded repeat ECMO runs within the same patient to avoid complexity and bias.We also excluded patients with missing blood pressure (systolic and diastolic at 24 hours of ECMO support) and cannulation information, central cannulation, on-ECMO percutaneous ventricular assist device or central venous access device support, coronary artery bypass graft or percutaneous coronary intervention, and post-cardiotomy shock.Patients with these conditions were excluded as they could impact the interpretation of PP readings and their association with ABI.

Data collection
The ELSO Registry collects ABG and hemodynamic information before and after ECMO cannulation (i.e., "pre-ECMO" and "on-ECMO", respectively).Pre-ECMO ABGs were drawn at maximum 6 hours before ECMO cannulation, and pre-ECMO ventilator settings were recorded within 6 hours of ECMO cannulation.If multiple ABGs existed within a speci c duration, the pre-ECMO ABG that was closest to the beginning of ECMO cannulation was selected.On-ECMO ABGs were drawn after ECMO cannulation started, no longer than 30 hours post-cannulation.If multiple ABGs were taken, the on-ECMO ABG nearest to 24 hours after the start of cannulation was chosen.On-ECMO hemodynamics were gathered closest to 24 hours after ECMO cannulation, though they could be collected at 18-30 hours after cannulation.Each variable was abstracted by a trained ELSO data manager/abstracter and was collected simultaneously.

De nitions
On-ECMO PP was calculated as "SBP at 24 hours" -"DBP at 24 hours".Delta partial pressure of arterial carbon dioxide (PaCO 2 ) was calculated as "On-ECMO PaCO 2 at 24 hours" -"Pre-ECMO PaCO 2 ".Pre-ECMO ventilator settings included conventional ventilation, high-frequency oscillatory ventilation, other high frequency ventilation (high frequency jet ventilation or percussive ventilation), other non-speci ed ventilations, and absence of ventilation.Pre-ECMO additional temporary mechanical circulatory support (tMCS) included intra-aortic balloon pump (IABP), Impella®, and left and right ventricular assist devices (though patients supported with on-ECMO tMCS were excluded from the analysis, as previously described).Pre-ECMO vasopressor infusions included dopamine, epinephrine, norepinephrine, phenylephrine, and vasopressin.Infusions were treated as a binary variable, meaning we treated them as the presence or absence of the infusions.Pre-ECMO vasopressor infusions were utilized for at least 6 hours within 24 hours of the start of ECMO cannulation.Pre-ECMO cardiac arrest was de ned as an event that required the use of cardiopulmonary resuscitation in conjunction with the administration of external cardiac massage within 24 hours of ECMO cannulation.Central cannulation was de ned as placement of the reinfusion cannula directly into the aorta.Peripheral cannulation was de ned as placement of cannula in a site other than the aorta (peripheral vessels).
On-ECMO complications included cardiac arrhythmia, hemolysis, renal replacement therapy, gastrointestinal hemorrhage, and ECMO circuit failure.De nitions for each complication are in the Supplemental Methods.ABI was de ned as the presence of central nervous system (CNS) infarction (ischemic stroke), diffuse ischemia (hypoxic-ischemic brain injury, HIBI), intra/extra parenchymal hemorrhage, intraventricular hemorrhage, seizures determined by electroencephalograph or clinically, and neurosurgical intervention (examples include intracranial pressure monitor, external ventricular drain, and craniotomy) and brain death during ECMO support.CNS ischemia was de ned as ischemic stroke (determined by ultrasound, computed tomography, CT, or magnetic resonance imaging, MRI) and HIBI (determined by CT or MRI).ICH included intra/extra parenchymal hemorrhage and intraventricular hemorrhage (both determined by CT or MRI).

Outcomes
The primary outcome was ABI during ECMO support between patients with PP ≤ 10 vs. PP > 10 mm Hg.The secondary outcomes were subtypes of ABI, CNS ischemia and ICH.

Statistical Analysis
Continuous variables were represented as median with interquartile range (IQR).Categorical variables were presented as frequency with percentages.The Wilcoxon rank-sum and Pearson's chi-square tests were utilized to compare continuous and categorical variables, respectively.Differences in PP between those with ABI vs. those without ABI were compared with Wilcoxon rank-sum.Statistical signi cance was set at a p-value < 0.05.Data missingness was handled with multiple imputation with ve separately imputed datasets (Rubin's Rules) 16 to augment statistical power.Continuous, unordered categorical, and dichotomous missing variables were imputed using regression with predictive mean matching, polytomous logistic regression, and logistic regression.All missing variables are shown in Supplemental Table 1.
Cubic spline analysis was utilized to non-linearly model the impact of PP on ABI.Based on in ection points ("spline knots") in this model, combined with prior data and clinical knowledge, we determined an appropriate PP threshold (≤ 10 mmHg) for logistic regression analysis.Boxplots were used to descriptively portray the association between PP vs. ABI.We performed univariable and multivariable logistic regression for ABI, CNS ischemia, and ICH in peripherally cannulated patients to determine if PP ≤ 10 mmHg was a signi cant risk factor for each of these outcomes even after adjustment for clinically relevant covariates.We chose covariates selected a priori based on clinical judgement and prior data for each model. 5,17 djusted covariates in the ABI model included age, sex, body mass index, pre-ECMO variables (additional tMCS, vasopressor infusions, cardiac arrest), on-ECMO variables (pH, arterial partial pressure of oxygen, PaO 2 ), delta PaCO 2 , and on-ECMO complications (hemolysis, arrhythmia, renal replacement therapy).In the CNS ischemia (ischemic stroke or HIBI) model, age, sex, pre-ECMO variables (PaCO 2 , PaO 2 , pH, vasopressor infusions, cardiac arrest), delta PaCO 2 , and on-ECMO complications (ECMO circuit failure, arrhythmia) were included in the adjustment.In the ICH model, age, sex, pre-ECMO variables (additional tMCS, vasopressor infusions, cardiac arrest), on-ECMO variables (PaO 2 , pH), delta PaCO 2 , and on-ECMO complications (ECMO circuit failure, gastrointestinal hemorrhage, hemolysis, renal replacement therapy) were included in the adjustment.In an exploratory analysis, we performed a multivariable logistic regression model for central VA-ECMO patients to determine if PP was associated with ABI.Adjusted odds ratios (aOR) were presented with 95% con dence intervals (CIs).All statistical analyses were performed using R Studio (R 4.1.2,www.r-project.org).

Discussion
In this ELSO Registry analysis of 9,807 peripheral VA-ECMO patients, we found that a low PP (≤ 10 mmHg) measured at 24-hours of ECMO support was independently associated with greater occurrence of ABI in peripheral VA-ECMO patients after adjusting for pre-selected clinically relevant covariates.Additionally, low PP was associated with CNS ischemia but not ICH.We also identi ed other risk factors for ABI in peripheral VA-ECMO patients: higher delta PaCO 2 , pre-ECMO cardiac arrest, and on-ECMO hemolysis, cardiac arrhythmia, and renal replacement therapy.
There may be multiple mechanisms at play that can lead to PP in uencing ABI in this cohort.2][23] These factors have been associated with increased incidence of ABI [24][25][26][27][28][29] in non-ECMO patients.Additionally, patients undergoing coronary artery bypass graft with IABP and cardiopulmonary bypass (i.e., nonpulsatile ow) have been shown to have less endothelial activation [30][31][32][33] and to observe a reduction in nitric oxide 34,35 due to systemic in ammatory response syndrome. 9,36,37 Ptients on the ECMO circuit frequently experience extreme changes in hemodynamic parameters such as PaO 2 17 and PaCO 2 , 38-40 which were previously shown to be associated with ABI.These blood gas derangements, combined with already compromised endothelium function and nonpulsatile cerebral blood ow, 41 may lead to ABI.Additionally, as left ventricular (LV) venting may lower PP and further predispose VA-ECMO patients to ABI, future research is warranted to investigate the effects of tMCS such as IABP and Impella® on the association between PP and ABI.
Our results demonstrated that low PP was associated with CNS ischemia but not ICH in peripheral VA-ECMO.Induced by ECMO circuit, the absence of pulsatility is associated with reduced O 2 consumption and impaired cerebral autoregulation, potentially contributing to CNS ischemia. 21Furthermore, low PP can indicate inadequate cardiac contractility, and thus systemic hypoperfusion, which may increase the risk of CNS ischemia.Interestingly, in contrast to previous literature describing that a larger delta PaCO 2 was associated with ICH, 42 our study demonstrated that delta PaCO 2 was associated with CNS ischemia rather than ICH.These results did not persist in central VA-ECMO which may be due to different hemodynamic states between both cohorts. 13Other additional key factors such as the use of systemic anticoagulation, duration of ECMO support, hemolysis, and platelet imbalance may also be involved in the pathophysiology of ICH. 43Overall, these ndings suggest that additional research is necessary to clarify how certain risk factors lead to either CNS ischemia or ICH in peripheral VA-ECMO patients.
Interestingly, unlike a previous study suggesting severe hyperoxia is associated with ABI, 17 an increase in PaO 2 was a protective factor for ABI in our analysis.One explanation is that aggressive oxygen therapy helps mitigate the effects of potential hypoxemia during ECMO support as hypoxemia in ECMO patients is well documented. 44,45 urthermore, a higher PaO 2 may help increase regional brain oxygen tension 46 and accordingly improve the cerebral metabolic rate of oxygen.We also note we did not "bin" PaO 2 values (continuous variable) by groups as this previous study did which may increase their risk of bias. 47verall, these results suggest that additional research with methodically rigorous study design is necessary to elucidate the mechanisms regarding how these physiological variables lead to ABI in continuous blood ow under the ECMO circuit.
Early assessment and recognition of myocardial function using low PP during the rst 24 hours of ECMO support has important clinical implications.Recognizing low PP may allow providers to promptly develop appropriate management techniques, including ne-tuning ECMO settings and using inotropes/vasopressors, 48 LV venting, 49 or pulsatile ECMO ow 50 to improve hemodynamics.Notably, our analysis showed that the use of vasopressors before ECMO support was protective of ABI, supporting our speculation.Additionally, upon recognizing low PP, clinicians can consider more vigilant and standardized neuromonitoring strategies to ensure adequate cerebral perfusion and prevent occurrence/worsening of ABI, which is especially important in peripheral VA-ECMO due to the potential for differential oxygenation. 51

Limitations
Our analysis was retrospective and observational, thus limiting our ability to determine causation effects.Additionally, the ELSO Registry lacks granular ABG data, only allowing us to extract one pre-ECMO and one on-ECMO data point for each patient in our analysis.Similarly, for PP we are limited to a single value at 24 hours, however, it is unclear if there is signi cant variance over the rst 24 hours that would in uence interpretation.This methodology has also been previously validated in an ELSO Registry analysis of PP and mortality in 2,400 VA-ECMO patients. 52Furthermore, it is unclear how such variance would systematically bias our results as any variation should be at random and thus favor the null hypothesis.Additionally, low PP may represent a population with higher severity of illness which may explain the higher occurrence of ABI despite attempting to account for this with statistical modeling.The ELSO Registry also does not contain speci c anticoagulation data, which is a known risk factor for ABI.Nevertheless, we adjusted for many ECMO-speci c and clinically relevant covariates in our analysis, and low PP was still independently associated with ABI.Additionally, our study represents the largest and most comprehensive analysis to-date investigating the association between PP and ABI in VA-ECMO patients.We also used methodically rigorous methods in our analysis including multiple imputation to handle missing data, thus minimizing bias and invigorating the validity of our analysis 53 and cubic spline analysis when identifying a PP threshold to abate the loss of information and poor predictions when using continuous variables. 54We also excluded ECMO patients simultaneously on LV venting devices as LV venting can directly modulate the PP in ECMO and thus could potentially confounding our ndings. 3inally, the optimum ECMO pump ow rate based on body surface area for each patient was not able to be determined in the ELSO Registry and should be noted.

Conclusions
In Missing values were handled with multiple imputations to increase statistical power.Pre-ECMO temporary mechanical circulatory support consisted of an intra-aortic balloon pump, Impella, and left ventricular assist devices.Pre-ECMO vasopressor infusions included dopamine, epinephrine, norepinephrine, phenylephrine, and vasopressin.Hemolysis was de ned as a peak plasma hemoglobin of at least 50 mg/dL occurring at least once during the ECMO run and sustained for at least 2 consecutive days.aOR: adjusted odds ratio CI: con dence interval.ECMO: extracorporeal membrane oxygenation.Forest plot of multivariable logistic regression model for occurrence of central nervous system ischemia in peripheral venoarterial extracorporeal membrane oxygenation patients.

Figure 3 Forest
Figure 3

Table 1 .
the largest analysis to-date of peripheral VA-ECMO patients with CS, a PP reading of 10 mmHg or less at the 24-hour time point of ECMO support was associated with increased occurrence of ABI.Low early PP was also uniquely associated with CNS ischemia but not ICH.Accordingly, PP during ECMO support may serve as a distinct marker for ABI in this high-risk population.Given these ndings, prospective observational studies investigating the association between PP and ABI with granular data and standardized neurological diagnoses is warranted.Baseline characteristics and clinical variables of venoarterial extracorporeal membrane oxygenation patients with cardiogenic shock strati ed by pulse pressure.