Pediatric cardiac extracorporeal membrane oxygenation: outcomes and risk factors from a multicenter study
Article information
Abstract
Background
Extracorporeal membrane oxygenation (ECMO) is a critical intervention for pediatric patients with cardiac-related medical indications. However, data on its outcomes and associated risk factors, particularly in Korea, are limited. This study aimed to evaluate these outcomes and identify predictors of mortality.
Methods
A multicenter retrospective study was conducted on 152 pediatric patients who underwent ECMO for cardiac-related medical conditions between 2012 and 2021. Data on patient demographics, clinical features, laboratory findings, and outcomes were collected and analyzed. Statistical methods included descriptive analysis, group comparisons, and regression analysis to evaluate risk factors for mortality.
Results
Cardiomyopathy was the most common indication for ECMO, and peripheral cannulation was used in 57.2% of cases. Acute kidney injury requiring dialysis was observed in 57.2% of patients. Non-extracorporeal cardiopulmonary resuscitation (ECPR) patients showed better outcomes, with a 68.4% survival-to-discharge rate compared to 33.3% in the ECPR group. Significant predictors of mortality included elevated lactate levels before ECMO, reduced 24-hour urine output post-ECMO, dialysis use, and pulseless electrical activity/asystole. While pre-ECMO left ventricular ejection fraction showed no overall association with outcomes, it was significantly higher in survivors with acute myocarditis.
Conclusion
This study provides valuable insights into the outcomes and risk factors of pediatric ECMO for cardiac-related medical indications in Korea. Lactate levels, urine output, and pre-ECMO rhythm were identified as prognostic factors. These results are expected to aid clinicians in managing this patient population and lay the groundwork for future prospective research.
INTRODUCTION
Since the first implementation of extracorporeal membrane oxygenation (ECMO) in humans during the early 1970s, the technology has advanced significantly. These advancements have enabled its successful application in pediatric patients, including neonates. The development and outcomes of ECMO have been systematically documented through the Extracorporeal Life Support Organization (ELSO) registry, providing valuable insights into the global use of extracorporeal life support (ECLS) [1].
However, the clinical characteristics and outcomes of ECMO in Korea remain largely unknown due to the limited participation of Korean centers in the ELSO registry. Furthermore, national or multicenter studies focusing on pediatric ECMO in Korea have not been conducted until recently. To bridge this knowledge gap, the Korean Society of Pediatric Critical Care Medicine funded a multicenter study to comprehensively evaluate the status of pediatric ECMO in Korea. The findings of this study, which provided an overview of patient volume and survival trends over the past decade, were published as a broad analysis of pediatric ECMO outcomes [2].
Building upon the foundational insights from the previous study, this research focuses specifically on pediatric cardiac patients with non-surgical conditions who were treated with ECMO. By narrowing the scope to this specific patient population, we aimed to evaluate the outcomes in detail and identify key risk factors associated with ECMO use in this subgroup.
METHODS
This study was approved by the Institutional Review Board of Pusan National University Yangsan Hospital (IRB No. 04-2022-004), Sejong General Hospital (IRB No. 2022-02-003), and the IRBs of all participating centers. The requirement for informed consent was waived in accordance with the policies governing retrospective studies.
Study Design
This multicenter retrospective study was conducted to collect and analyze data on pediatric patients who underwent ECMO insertion in Korea. We reviewed the medical records of all patients who received ECMO for any indication between January 2012 and December 2021, as detailed in the initial report [2]. A total of 10 centers that possessed eligible cardiac-related medical ECMO cases and agreed to participate in this subgroup analysis were included in the study.
Data Collection
The methodology for data collection has been described in detail in the previous report [2]. Pediatric patients under 18 years of age were included in the study. Data were collected on demographics, clinical characteristics, and outcomes. The data included laboratory results, echocardiographic findings, heart rhythm data, urine output, and other relevant clinical parameters. For this subgroup analysis, we focused on patients with severe heart failure necessitating ECMO support, excluding those in a post-cardiotomy state. This patient group, referred to as the "cardiac-related medical group," included conditions such as acute myocarditis, cardiomyopathy, congenital heart disease, and others (e.g., malignant arrhythmia).
We also paid particular attention to the application of extracorporeal cardiopulmonary resuscitation (ECPR), which was defined according to the ELSO guidelines as the initiation of ECMO during conventional cardiopulmonary resuscitation with manual or mechanical chest compressions, or within 20 minutes of return of spontaneous circulation without ongoing compressions [3].
Outcome Measures
While the previous report provided an overview of individual outcomes, this study specifically targeted the clinical characteristics of the cardiac-related medical group and their association with in-hospital outcomes. The primary in-hospital outcomes assessed were survival to discharge and death. Patients were classified as having survived to discharge if they were discharged alive from the hospital, irrespective of whether they were discharged to home or transferred to another facility. However, transfers while still on ECMO support were not classified as survival to discharge. ECMO weaning was considered successful if the patient survived for at least 48 hours after ECMO removal.
We explored potential correlations between ECMO outcomes and various clinical features, including laboratory parameters (e.g., lactate, B-type natriuretic peptide [BNP], N-terminal prohormone of brain natriuretic peptide [NT-proBNP], creatine kinase-MB [CK-MB]), which were measured immediately before ECMO insertion or on the same day; 24-hour urine output following ECMO initiation; the use of dialysis; and patient information immediately prior to ECMO insertion, such as left ventricular ejection fraction (LVEF) and heart rhythm.
Statistical Analysis
Data were analyzed using IBM SPSS statistics for Windows version 30 (IBM Corp.). Descriptive statistics were presented as medians with interquartile ranges for continuous variables or frequencies with percentages for categorical variables. Group comparisons were performed using the chi-square test for nominal variables and the Mann-Whitney U-test for continuous variables, as most data did not follow a normal distribution. Binary logistic regression analysis was conducted to examine the association between pre-ECMO rhythms and mortality. The independent variable was the rhythm type at the time of ECMO insertion, classified into seven predefined groups. The dependent variable was mortality. Odds ratios (ORs) with 95% CIs were calculated to estimate the likelihood of mortality associated with each rhythm group, with normal sinus rhythm serving as the reference category.
RESULTS
In total, 152 patients were enrolled in the study. Their demographic and clinical characteristics are summarized in Table 1. The median age of the patients was 4.36 years. Among the cardiac-related medical indication group, cardiomyopathy was the most common diagnostic category. Peripheral cannulation, such as via the jugular artery and vein, was more frequently performed (57.2%) than the central approach via sternotomy. One-third of the patients underwent ECMO insertion during cardiopulmonary resuscitation (ECPR), with a median time from the initiation of chest compressions to the establishment of ECMO flow of 70 minutes.
Demographic and clinical features of the patients
Patients were supported by ECMO for a median of 8.5 days. Among them, 10% transitioned to ventricular assist device (VAD) support, which extended the total ECLS duration (including both ECMO and VAD support) by a median of 11.2 days for temporary VADs or 169 days for durable VADs, depending on the type of VAD used. Dialysis therapies, including continuous renal replacement therapy (CRRT) and peritoneal dialysis, were implemented during hospitalization in half of the patients, with CRRT being the predominant modality (Table 1).
In-Hospital Outcomes
The in-hospital outcomes of patients with cardiac-related medical conditions are shown in Table 2. More than three-quarters of non-ECPR patients were successfully weaned from ECMO, and 68% of them survived to discharge. In contrast, ECPR patients exhibited worse in-hospital outcomes, as expected, with higher mortality rates and lower survival to discharge. However, the number and distribution of heart transplantations did not differ significantly between the non-ECPR and ECPR groups. Among patients awaiting transplantation, one of three patients in the non-ECPR group and the only patient in the ECPR group died while on the waiting list.
In-hospital outcomes of cardiac-related medical ECMO patients
Associations with Mortality
Various clinical features were analyzed for their association with mortality. Non-survivors exhibited higher lactate levels before ECMO insertion, lower 24-hour urine output immediately after ECMO initiation, and a higher prevalence of dialysis use (Table 3, Fig. 1A and B). However, other parameters, such as BNP, NT-proBNP, CK-MB, and LVEF prior to ECMO insertion, did not show statistically significant differences between survivors and non-survivors. Additionally, the timing and duration of dialysis were not significantly associated with mortality (Table 3, Fig. 1C-F). Notably, in a subgroup analysis of patients diagnosed with acute myocarditis, pre-ECMO LVEF was significantly higher in survivors than in non-survivors (median LVEF, 18.15% in non-survivors vs. 29.86% in survivors, p=0.043) (Supplementary Table 1).
Clinical features of cardiac-related medical ECMO patients
The distribution of clinical features, expressed as boxplots based on in-hospital outcomes. (A) Lactate levels prior to extracorporeal membrane oxygenation (ECMO) initiation showed a statistically significant difference between survivors and non-survivors (p=0.001). (B) The 24-hour urine output following ECMO initiation was significantly higher in survivors than in non-survivors (p=0.037). (C) B-type natriuretic peptide (BNP) levels prior to ECMO initiation. (D) N-terminal prohormone of brain natriuretic peptide (NT-proBNP) levels prior to ECMO initiation. (E) Creatine kinase-MB (CK-MB) levels prior to ECMO initiation. (F) Left ventricular ejection fraction (LVEF) prior to ECMO initiation. No statistically significant differences were observed for clinical features (C-F) between the groups.
The distribution of heart rhythms immediately prior to ECMO insertion is shown in Fig. 2, categorized by in-hospital outcomes. The heart rhythms were classified into seven groups: ventricular tachycardia (VT)/ventricular fibrillation (VF), atrial fibrillation/atrial flutter, high-degree atrioventricular (AV) block, sinus bradycardia (heart rate <60 bpm), pulseless electrical activity (PEA)/asystole, normal sinus rhythm, and others. Normal sinus rhythm was the most common rhythm at the time of ECMO insertion, followed by VT/VF (Fig. 2). Logistic regression analysis revealed that PEA/asystole was associated with a significantly higher likelihood of mortality, with a 9.6-fold increased risk compared to normal sinus rhythm (Table 4). Other heart rhythms, however, were not significantly associated with mortality.
Heart rhythms prior to extracorporeal membrane oxygenation initiation, presented using a 100% stacked bar chart, categorized by in-hospital outcomes. Normal sinus rhythm was the most prevalent heart rhythm, followed by ventricular tachycardia (VT)/ventricular fibrillation (VF). Pulseless electrical activity (PEA)/asystole was identified as the rhythm with the worst prognosis (odds ratio, 9.6; p<0.001). AF, atrial fibrillation; AFL, atrial flutter; AV, atrioventricular; HR, heart rate.
The association between heart rhythm before ECMO insertion and mortality
DISCUSSION
Since the first pediatric ECMO cases, including one case where the patient survived, were documented in Korea in 1991, the usage and outcomes of ECMO have steadily improved [4]. However, no systematic studies have comprehensively investigated current practices and outcomes in pediatric ECMO. Our multicenter study was designed to address this gap. The previous report provided an overview of pediatric ECMO practices in Korea [2], while the present study extends this effort by focusing on cardiac-related medical indications and exploring factors influencing ECMO outcomes.
Over the 10-year study period, 152 ECMO runs for cardiac-related medical indications were performed across 10 participating centers. The sex distribution was balanced across the cohort, and cardiomyopathy, including its subtypes, was the most common diagnosis (Table 1). Peripheral cannulation was used more frequently than central cannulation (57.2%), which may be attributed to the fact that sternotomy is not easy or feasible in emergency situations, unlike post-cardiotomy cases where the sternum has been already opened or can be accessed during surgery. Transition to VAD occurred in 10.5% of cases (temporary VAD, in 7.9%; durable VAD, in 2.6%), which is lower than the rate of 23.5% reported in another study [5], likely due to the approval for national insurance reimbursement for VADs only in 2018.
It is also noticeable that acute kidney injury requiring dialysis developed in more than half of the patients (57.2%), as some previous studies have reported that acute kidney injury and renal replacement therapy were both associated with a longer duration of ECMO support and an increased risk of mortality at hospital discharge [6,7]. In fact, the proportion of patients who were treated with renal replacement therapy in this study (57.2%) was slightly higher than that reported previously (48.4%) [6]. This might contribute to the somewhat worse outcomes compared to the ELSO registry data described below.
According to our previous report, ECMO was more frequently used for post-cardiotomy indications than for cardiac-related medical indications. However, in-hospital outcomes, as represented by the rate of survival to discharge, were better in cardiac-related medical cases (36.5% in post-cardiotomy cases vs. 51.2% in cardiac-related medical cases) [2]. While pulmonary hypertension and sepsis were included as cardiac-related medical indications in the previous report, the current study excluded these conditions. Despite this difference in inclusion criteria, the rate of survival to discharge in this study was comparable to that in the previous report and to outcomes reported for cardiac-related medical patients in the 2016 ELSO registry (55.9% in the current study vs. 51.2% in the previous report vs. 60% in the ELSO registry report of 2016) (Table 2) [8].
Additionally, we observed that patients who underwent ECPR exhibited worse outcomes than those who did not (survived to discharge, 68.4% in the non-ECPR group vs. 33.3% in the ECPR group). This aligns with findings from the ELSO registry, which also reported lower survival rates in the ECPR group than in other pediatric ECMO groups, although the survival rate in the ELSO registry was higher than that observed in our cohort (41%–43% in the ELSO registry vs. 33.3% in our cohort) [1,8]. These findings are consistent with expectations; however, previous studies have shown conflicting results regarding whether ECPR is an independent risk factor for mortality [7,9]. The worse outcomes in our ECPR group may be attributable to the prolonged time from the onset of chest compressions to the establishment of ECMO flow, with a median duration of 70 minutes (Table 1). Recent studies have highlighted that longer intervals between chest compression initiation and ECMO flow are associated with increased mortality [10,11].
Risk Factors for Mortality
Our analysis confirmed several previously identified risk factors for mortality. Non-survivors had higher lactate levels before ECMO insertion, lower urine output within 24 hours after initiation, and a higher prevalence of dialysis (Table 3). These findings corroborate previous studies highlighting the prognostic value of lactate levels, acute kidney injury, and poor urine output [5-7,9]. Additionally, the rate of lactate clearance has been identified as a dynamic predictor of mortality, suggesting that both static and dynamic lactate metrics are critical for prognosis [12]. However, it is important to acknowledge that elevated lactate levels might also reflect poor pre-ECMO clinical status, potentially due to delayed ECMO initiation, which could lead to further lactate accumulation. In this context, our findings underscore the potential importance of early ECMO initiation to mitigate the progression of metabolic derangement and improve patient outcomes.
However, traditional cardiac markers such as BNP, NT-proBNP, CK-MB, and pre-ECMO LVEF did not show significant predictive value in this study. While this may suggest that these markers are not reliable risk factors in the context of pediatric ECMO for cardiac-related medical indications, it is also possible that this result reflects the limitations of missing data or patient heterogeneity. In particular, the lack of simultaneous testing for BNP and NT-proBNP across most centers resulted in insufficient data to draw reliable conclusions. Interestingly, when analyzing a relatively homogeneous patient group with acute myocarditis separately, pre-ECMO LVEF was found to have a significant association with mortality. This result is consistent with findings from a previous study. Tuan et al. [13] identified LVEF at 48 hours, platelet count, and protein levels as predictors of mortality in patients with acute myocarditis. Their study also analyzed patients receiving ECMO and highlighted LVEF at 48 hours, a high vasoactive-inotropic score, and lactate at 24 hours post-ECMO as significant risk factors. In contrast, a study by Sun et al. [14] on children with acute myocarditis receiving venoarterial ECMO identified peak serum creatinine and peak CK-MB during ECMO as prognostic factors, unlike our findings. These discrepancies may stem from differences in study design, patient populations, or clinical practices, and underscore the need for further research to confirm these associations.
Heart Rhythm and Mortality
Heart rhythm immediately before ECMO insertion showed a significant association with in-hospital outcomes. Normal sinus rhythm was more prevalent in survivors, whereas PEA/asystole was predominant in non-survivors (Fig. 2). Logistic regression analysis revealed that PEA/asystole was associated with a 9.6-fold higher likelihood of mortality compared to normal sinus rhythm (Table 4). Other rhythms, including VT/VF, high-degree AV block, sinus bradycardia, and others, were not significantly associated with mortality. Few studies have evaluated pre-ECMO heart rhythm as a prognostic factor in pediatric patients, but findings from adult studies support the protective role of normal sinus rhythm before ECMO initiation (OR, 0.374) [15]. Our results suggest that PEA/asystole remains a critical risk factor for mortality, while other rhythms may not independently predict outcomes.
Study Limitations
This study has several limitations. First, data loss due to the emergency nature of ECMO procedures and the retrospective study design may have affected the reliability of certain parameters, such as lactate, CK-MB, BNP, NT-proBNP, and LVEF. Additionally, analysis for troponin I and troponin T could not be performed due to a high proportion of missing data. However, the consistency of lactate levels as a prognostic factor suggests that the most critical predictors were not overlooked. Second, since not all major centers participated in this study on ECMO for pediatric patients with cardiac-related medical conditions, selection bias may exist, and the study results may not fully represent current ECMO practices in Korea. Therefore, the generalizability of our findings is limited.
Conclusions
Building upon the recent study that first reported on the practice and outcomes of pediatric ECMO in Korea, our study provides focused insights into the outcomes and risk factors specific to pediatric cardiac-related medical ECMO patients. The results are consistent with findings from previous studies, identifying the pre-ECMO lactate level, 24-hour urine output post-ECMO, use of dialysis, and PEA/asystole as significant risk predictors. This study is expected to offer clinicians valuable insights into managing pediatric ECMO patients with cardiac-related medical indications, aiding in more accurate prognostic predictions and improved decision-making. Furthermore, it serves as a foundational step toward future prospective studies to further advance the understanding and management of this patient population.
Notes
CONFLICT OF INTEREST
Won Kyoung Jhang is an editor-in-chief, and Jung Eun Kwon, Bongjin Lee, Joongbum Cho, and Younga Kim are editorial board members of the journal but were not involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.
FUNDING
This study was supported by a 2022 research grant from the Korean Society of Pediatric Critical Care Medicine.
AUTHOR CONTRIBUTIONS
Conceptualization: HC, JYN, JC, HJC, JHB, YK. Data curation: HC, MO, JEK, JAS, KHL, WKJ, BL, AYC, JC, HJC, JHB. Supervision: YK. Project administration: YK. Funding acquisition: WKJ, YK. Writing - original draft: HC. Writing - review & editing: YK. All authors read and agreed to the published version of the manuscript.
SUPPLEMENTARY MATERIALS
Supplementary materials can be found via https://doi.org/10.32990/apcc.2024.00157.
Clinical features of acute myocarditis subgroup