Novel Risk Scoring Model to Predict the Implementation of Veno-Arterial Extracorporeal Membrane Oxygenation in Patients With Acute Myocarditis

Abstract

Background: This study aimed to identify risk factors associated with the implementation of veno-arterial extracorporeal membrane oxygenation (VA-ECMO) in patients with acute myocarditis and to develop a predictive model.

Methods and Results: This retrospective study included 841 patients from 7 hospitals in Korea with biopsy-proven or clinically suspected acute myocarditis. Logistic regression analysis was used to identify the clinical characteristics of patients who required VA-ECMO and to construct a scoring system to predict the implementation of VA-ECMO. Among the study population, 217 (25.8%) patients underwent VA-ECMO. The study population was divided into training (n=621) and testing (n=220) cohorts according to participating center. The final predictive model of VA-ECMO insertion derived from the training cohort included the following: initial mean blood pressure <65 mmHg, cardiac arrest, Glasgow Coma Scale score ≤12, platelet count <100×10³/mL, pulmonary congestion on chest X-ray, QRS interval ≥120 ms, left or right bundle branch block, and left ventricular ejection fraction <40%. Using this predictive model, a β coefficient-weighted Korean Acute Myocarditis (KAM) score was developed. External validation of the predictive model and KAM score using the testing cohort showed excellent discriminant ability (areas under the curve of 0.945 and 0.921, respectively).

Conclusions: A risk scoring system based on simple clinical and laboratory parameters at initial presentation could predict the implementation of VA-ECMO and clinical course in patients with acute myocarditis.

The clinical course of acute myocarditis can vary widely, ranging from mild chest pain or dyspnea without affecting hemodynamics to life-threatening arrhythmias and heart failure with severe cardiogenic shock (CS).^1,2 In uncomplicated cases, a diagnosis can be made without endomyocardial biopsy and conservative management is sufficient.² Conversely, complicated cases may require endocardial biopsy to confirm diagnosis and guide etiologically specific treatment, as well as mechanical circulatory support or pacemaker implantation for treatment.²

The challenges in managing acute myocarditis are that it is rare and its clinical presentation is non-specific and varies widely.¹ More importantly, it is difficult to predict the clinical course of an individual patient, which is problematic given that patients with acute myocarditis can deteriorate rapidly.^3,4 Recently, several studies have shown that patients with a complicated and fulminant course have worse short- and long-term prognoses than those with non-fulminant disease.^5–7 These studies also evaluated associations between clinical features and progression to fulminant disease, but included only a limited number of patients with acute myocarditis. Such data are critical to accurately predict which patients will progress to refractory CS that would eventually require veno-arterial extracorporeal membrane oxygenation (VA-ECMO).

Therefore, the present study used a dedicated multicenter registry to identify factors associated with the implementation of VA-ECMO in patients with acute myocarditis. The study also aimed to develop a simple scoring system capable of predicting the implementation of VA-ECMO in real-world practice.

Methods

Study Design and Population

The study population for this multicenter observational study was retrospectively gathered from 7 hospitals (Samsung Medical Center, Chonnam National University Hospital, Asan Medical Center, Keimyung University Dongsan Hospital, Bucheon Sejong Hospital, Korea University Anam Hospital, Incheon Sejong Hospital) in Korea from January 2004 to December 2022. The study was registered with ClinicalTrials.gov (Multicenter Retrospective Observational Study of Acute Myocarditis in Korea; ID: NCT05933902). Patients who met the diagnostic criteria of clinically suspected or biopsy-proven acute myocarditis according to the European Society of Cardiology position statement were included.⁸ Briefly, a diagnosis of clinically suspected myocarditis was defined as the identification of ≥1 clinical presentation and ≥1 diagnostic criterion. The relevant clinical presentations were: (1) acute pericarditic or pseudo-ischemic chest pain; (2) new-onset (within days to 3 months) or worsening dyspnea at rest or exercise; (3) palpitations, unexplained arrhythmia symptoms, syncope, and/or aborted sudden cardiac death; and (4) unexplained CS. The relevant diagnostic criteria were: (1) newly abnormal 12-lead electrocardiography results; (2) elevated troponin T or I; (3) functional and structural abnormalities on echocardiography or cardiac magnetic resonance imaging; and (4) edema and/or late gadolinium enhancement of the classical myocarditis pattern on cardiac magnetic resonance imaging.⁹ Biopsy-proven myocarditis was defined in accordance with the Dallas criteria.¹⁰ Patients with alternative cardiac diagnoses at discharge, including ischemic heart disease and sarcoidosis, were excluded from the study.

In this study, the study population was divided into training and testing cohorts according to the participating center: patients from 6 centers other than Samsung Medical Center were allocated to the training cohort and patients from Samsung Medical Center were allocated to the testing cohort (Supplementary Figure 1). Of note, only the training cohort was used to develop the predictive model and scoring system to predict the implementation of VA-ECMO. The testing cohort was not used in model training and was used only to validate the predictive model and scoring system built on the training dataset.

The study protocol was approved by the Institutional Review Board at Samsung Medical Center (Reference no. 2022-01-003) and the study was conducted in accordance with the principals of the Declaration of Helsinki. The Institutional Review Board waived the requirement for informed consent because patients were enrolled in the study retrospectively and data were collected after being anonymized.

Data Collection, Follow-up, and Clinical Outcomes

The following data were collected retrospectively by reviewing electronic medical records: demographic features, comorbidities, clinical presentations, laboratory findings, electrocardiography, echocardiography findings within 24 h after admission and before VA-ECMO implementation, cardiac magnetic resonance imaging, cardiac pathologic findings, and in-hospital management, including medications and organ-support device before VA-ECMO implementation during hospitalization. In patients who underwent mechanical ventilation (n=244; 29.0% of the total population), the lowest Glasgow Coma Scale score before starting ventilation was obtained. In addition, dedicated study coordinators sent queries to individual researchers by email or telephone to clarify and to fill in missing values in the collected data, if necessary. The primary outcome of this study was the implementation of VA-ECMO for refractory CS that did not respond to inotropes or vasopressors, or for cardiac arrest regardless of resuscitation. The secondary outcome was in-hospital mortality of any cause. Of note, the actual use of left ventricular assist devices became feasible in Korea only after the Korean Government approved their use in October 2018, which limited their inclusion as a clinical outcome in this study.

Statistical Analyses

Categorical variables are presented as numbers and relative frequencies (percentages) and were compared using the Chi-squared test. Continuous variables are presented as the mean±SD or as the median with interquartile range (IQR) and were compared using Student’s t-test or the Mann-Whitney rank-sum test, as appropriate. To identify independent predictors of the primary outcome, variables that were clinically relevant based on previous studies and significantly associated with the implementation of VA-ECMO with P<0.05 in the univariable analyses were considered as potential predictor variables.^5,6,11–14 Only binary variables were considered in the multivariable model to achieve simplicity in clinical application. Cut-off values for dichotomization of continuous variables were predetermined according to clinical relevance or were derived from receiver operating characteristic (ROC) curve analyses. Then, the final multivariable predictive model was constructed using backward elimination based on the Akaike information criterion to select the set of variables that make the best information criterion to predict the implementation of VA-ECMO. The effects of variables are presented as odds ratios (OR) and 95% confidence intervals (CI).

The discriminant ability of the constructed predictive model was assessed using ROC curve analysis, and the area under the curve (AUC) was calculated. For internal validation of the predictive model, K-fold cross-validation and bootstrapping were performed. First, K-fold cross-validation was performed by randomly dividing the training dataset into 10 equally sized sub-datasets. Nine sub-datasets were used for training, and the remaining sub-dataset was used for validation. This process was repeated 10 times to achieve accurate results, and each of the 10 sub-datasets was only used once for model validation. The mean value of these 10 calculations was used to assess the reliability of the model. Second, bootstrapping was performed by sampling the training dataset using 1,000 repetitions with replacement. Primarily, complete-case analyses were used, which only consider patients without missing values for the variables of interest. In addition, analyses using K-nearest neighbor imputation of missing values in variables of interest were also performed as sensitivity analyses.

Based on the predictive model, a simplified Korean Acute Myocarditis (KAM) score was developed using a β coefficient-based scoring method. To generate a simple integer-based point score for each selected predictive variable, each β coefficient was divided by the absolute value of the smallest β coefficient and rounded to the nearest integer. Model calibration was assessed using a Hosmer-Lemeshow goodness-of-fit analysis. External validation of the predictive model and KAM score was performed using the testing cohort.

The associations between the KAM score and estimated in-hospital mortality risk were plotted using a penalized spline curve with 3 degrees of freedom and were calculated using a logistic regression model. All probability values were 2-sided, and P<0.05 was considered statistically significant. All statistical analyses were performed using R version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Characteristics of the Study Population

In the study population, 217 (25.8%) patients underwent VA-ECMO, and 95 (11.3%) patients died during the index hospitalization. For those who underwent VA-ECMO, the median time from admission to VA-ECMO implementation was 1,309.5 min (IQR 768.3–2,091.0 min) and the time from shock to VA-ECMO implementation was 276.5 min (IQR 54.0–2,091.0 min).

The study population was divided into training (n=621) and testing (n=220) cohorts according to participating center (Supplementary Figure 1). The baseline characteristics of the training cohort are summarized in Table 1 and Supplementary Table 1. The mean age of patients in the training cohort was 36.1±20.6 years, and 37.2% were female. The most frequent clinical presentation was chest pain (65.7%), followed by febrile sense (51.2%) and dyspnea (48.6%). Nearly half the patients had pulmonary congestion on chest X-ray and the median serum N-terminal pro-B-type natriuretic peptide concentration was 2,622.5 pg/mL (IQR 356.0–13,438.0 pg/mL). Compared with patients without VA-ECMO, patients who underwent VA-ECMO were older and a higher proportion had dyspnea as a clinical presentation. Overall, patients with VA-ECMO had significantly worse vital signs, organ dysfunction indices, and laboratory and electrocardiographic findings than patients without VA-ECMO. Echocardiographic evaluation indicated that left ventricular ejection fraction (LVEF) was lower, and that E/e′ and right ventricular systolic pressure were higher in patients with than without VA-ECMO. Endomyocardial biopsy was performed in 131 (21.1%) patients and was performed more frequently in patients with than without VA-ECMO (Table 2). Regarding in-hospital management, diuretics and organ support were used more frequently in patients with than without VA-ECMO, whereas anti-inflammatory drugs were less frequently used in patients with VA-ECMO. In patients with VA-ECMO, the median ECMO duration was 7 days, and distal perfusion and left ventricular unloading were performed in approximately half the patients (Supplementary Table 2).

Table 1.

Baseline Characteristics of Acute Myocarditis Patients With or Without VA-ECMO

	No. patients with available data	Entire cohort (n=621)	With VA-ECMO (n=126)	Without VA-ECMO (n=495)	P value
Demographics
Age (years)	621	36.1±20.6	40.7±18.7	35.0±20.9	0.005
Female sex	621	231 (37.2)	56 (44.4)	175 (35.4)	0.075
BMI (kg/m2)	565	23.3±4.3	23.2±4.2	23.4±4.3	0.656
Comorbidities
Hypertension	621	94 (15.1)	22 (17.5)	72 (14.5)	0.499
Diabetes	621	38 (6.1)	7 (5.6)	31 (6.3)	0.930
Dyslipidemia	621	38 (6.1)	11 (8.7)	27 (5.5)	0.245
Chronic kidney disease	621	12 (1.9)	1 (0.8)	11 (2.2)	0.498
Current smoking	621	95 (15.3)	25 (19.8)	70 (14.1)	0.148
Malignancy	621	17 (2.7)	7 (5.6)	10 (2.0)	0.062
Previous history of myocarditis	621	8 (1.3)	0 (0)	8 (1.6)	0.320
Clinical presentation
Chest pain	621	408 (65.7)	68 (54.0)	340 (68.7)	0.003
Dyspnea	621	302 (48.6)	94 (74.6)	208 (42.0)	<0.001
Syncope	621	51 (8.2)	8 (6.3)	43 (8.7)	0.502
Respiratory symptoms other than dyspnea	621	130 (20.9)	38 (30.2)	92 (18.6)	0.006
Febrile sense	621	318 (51.2)	82 (65.1)	236 (47.7)	0.001
Mean blood pressure (mmHg)	617	69.0±18.0	53.3±18.1	72.8±15.8	<0.001
Bradycardia	617	53 (8.6)	15 (12.1)	38 (7.7)	0.168
Tachycardia	617	205 (33.2)	65 (52.4)	140 (28.4)	<0.001
Tachypnea	601	189 (31.4)	55 (45.8)	134 (27.9)	<0.001
Fever	621	148 (23.8)	49 (38.9)	99 (20.0)	<0.001
Cardiac arrest	621	64 (10.3)	52 (41.3)	12 (2.4)	<0.001
Arrest rhythm	52/64				0.125
Ventricular tachycardia or fibrillation		32/52 (61.5)	29/43 (67.4)	3/9 (33.3)
Pulseless electrical activity or asystole		20/52 (38.5)	14/43 (32.6)	6/9 (66.7)
Diagnosis methods	621				<0.001
Biopsy confirmed		102 (16.4)	37 (29.4)	65 (13.1)
Clinically suspected myocarditis		519 (83.6)	89 (70.6)	430 (86.9)
Vasoactive inotropic score	621	0 [0–10.0]	36.0 [13.0–68.0]	0.0 [0.0–3.0]	<0.001
PaO2/FiO2 ratio	374	332.0 [169.0–427.6]	180.0 [107.3–370.0]	371.4 [230.0–450.0]	<0.001
GCS	523				<0.001
Mild (13–15)		415 (79.3)	45 (36.9)	370 (92.3)
Moderate (9–12)		13 (2.5)	9 (7.4)	4 (1.0)
Severe (3–8)		95 (18.2)	68 (55.7)	27 (6.7)
SOFA score	510	2.0 [0–6.0]	9.0 [6.0–14.0]	0.0 [0–4.0]	<0.001

Unless indicated otherwise, data are given as the mean±SD, median [interquartile range], n (%), or n/N (%). BMI, body mass index; FiO₂, fractional inspired oxygen; GCS, Glasgow Coma Scale; PaO₂, arterial pressure of oxygen; SOFA, Sequential Organ Failure Assessment; VA-ECMO, veno-arterial extracorporeal membrane oxygenation.

Table 2.

Echocardiographic, Cardiac Magnetic Resonance Imaging, and Pathologic Findings

	No. patients with available data	Entire cohort (n=621)	With VA-ECMO (n=126)	Without VA-ECMO (n=495)	P value
Echocardiography
LV ejection fraction (%)	596	44.7±17.8	27.2±17.6	49.0±15.0	<0.001
LV end diastolic dimension (mm)	537	47.4±7.5	47.6±8.8	47.3±7.2	0.793
LV end systolic dimension (mm)	520	35.0±8.5	40.0±10.1	33.9±7.6	<0.001
Septal wall thickness (mm)	521	9.9±2.4	10.1±2.7	9.8±2.3	0.231
Posterior wall thickness (mm)	522	9.8±2.3	10.1±2.5	9.8±2.2	0.167
Left atrial volume index (mL/m2)	215	30.0±11.8	34.8±18.8	29.2±10.1	0.120
E velocity (m/s)	464	0.7±0.2	0.6±0.3	0.7±0.2	<0.001
A velocity (m/s)	351	0.6±0.3	0.5±0.2	0.6±0.4	0.029
e′ velocity (cm/s)	452	7.7±3.5	5.3±2.5	8.2±3.4	<0.001
E/e′	442	10.6±4.9	12.8±5.9	10.2±4.6	0.001
RV systolic pressure (mmHg)	378	25.9±9.8	22.4±9.9	26.5±9.7	0.004
Pericardial effusion	568	272 (47.9)	58 (54.7)	214 (46.3)	0.146
Cardiac magnetic resonance imaging
Edema	207	75 (36.2)	8 (38.1)	67 (36.0)	0.999
Late gadolinium enhancement	208	155 (74.5)	12 (57.1)	143 (76.5)	0.096
Pericardial effusion	206	104 (50.5)	9 (42.9)	95 (51.4)	0.612
Endomyocardial biopsy
Performed	621	131 (21.1)	42 (33.3)	89 (18.0)	<0.001
Histologic diagnosis	131/131				0.053
Lymphocytic		87/131 (66.4)	34/42 (81.0)	53/89 (59.6)
Eosinophilic		15/131 (11.5)	3/42 (7.1)	12/89 (13.5)
Giant cell		0/131 (0)	0/42 (0)	0/89 (0)
Non-diagnostic		29/131 (22.1)	5/42 (11.9)	24/89 (27.0)

Unless indicated otherwise, data are given as the mean±SD, n (%), or n/N (%). LV, left ventricular; RV, right ventricular; VA-ECMO, veno-arterial extracorporeal membrane oxygenation.

When comparing the training and testing cohorts, although the proportion of females was higher in the testing than training cohort, the demographic characteristics and comorbidities were mostly similar between the 2 groups. However, the severity of patients was higher in the testing cohort than the training cohort, as supported by N-terminal pro-B-type natriuretic peptide concentration, the proportion of patients with pulmonary congestion on chest X-ray, electrocardiographic findings, LVEF on echocardiography, the proportion of patients in whom cardiac arrest occurred, and organ dysfunction indices. Consequently, vasopressor or organ support were more frequently used in the testing than training cohort (Supplementary Tables 3–5).

Predictive Model for Implementation of VA-ECMO

The predictors of VA-ECMO implementation in the univariable and final multivariable logistic regression model are presented in Supplementary Table 6 and Table 3, respectively. The predictors significantly associated with VA-ECMO implementation as identified by backward selection were initial mean blood pressure <65 mmHg, cardiac arrest, Glasgow Coma Scale score ≤12, platelet count <100×10³/mL, pulmonary congestion on chest X-ray, QRS interval ≥120 ms, left or right bundle branch block, and LVEF <40%.

Table 3.

Independent Predictors of the Requirement for VA-ECMO

	Multivariable analysis		β coefficient
	OR (95% CI)	P value
Initial mean blood pressure <65 mmHg	1.702 (0.829–3.491)	0.147	0.532
Cardiac arrest	8.521 (3.487–20.819)	<0.001	2.142
GCS score ≤12	12.020 (5.638–25.624)	<0.001	2.487
Platelet count <100×103/mL	3.078 (1.001–9.465)	0.050	1.124
Pulmonary congestion on chest X-ray	6.030 (2.738–13.281)	<0.001	1.797
QRS interval ≥120 ms	3.421 (1.351–8.664)	0.010	1.230
Left or right bundle branch block	2.290 (0.901–5.816)	0.082	0.828
Ejection fraction <40%	2.810 (1.381–5.716)	0.004	1.033

CI, confidence interval; OR, odds ratio. Other abbreviations as in Table 1.

The AUC representing the discriminant ability of the predictive model was 0.945 (95% CI 0.921–0.969; Supplementary Figure 2A). Internal validation of the predictive model using K-fold cross-validation (AUC 0.940; 95% CI 0.901–0.979) or bootstrapping (AUC 0.936; 95% CI 0.897–0.975) methods showed similar discriminant ability (Supplementary Figure 2B,C). In sensitivity analyses using K-nearest neighbor imputation, consistent results were found (Supplementary Figure 3).

KAM Score to Predict Implementation of VA-ECMO

For simplicity and clinical applicability, the KAM scoring system was developed based on the final multivariable logistic regression model to predict the implementation of VA-ECMO. The variables and their assigned scores were as follows: 1 point for initial mean blood pressure <65 mmHg; 4 points for cardiac arrest; 5 points for Glasgow Coma Scale score ≤12; 2 points for platelet count <100×10³/mL; 3 points for pulmonary congestion on chest X-ray; 2 points for QRS interval ≥120 ms; 2 points for left or right bundle branch block; and 2 points for LVEF <40%. The total KAM score ranged from 0 to 21 (Figure 1A). The distribution of the training cohort patients according to their KAM scores is shown in Figure 1B. When predicting the implementation of VA-ECMO, the optimal cut-off value of the KAM score was 8. The AUC, sensitivity, and specificity of the KAM score were 0.945 (95% CI 0.921–0.969), 90.7%, and 88.7%, respectively. Furthermore, the negative predictive value of the KAM score was 97.1% (Figure 1C). The VA-ECMO implementation rate increased significantly as the KAM score increased (OR 1.683 for every 1-point increase; 95% CI 1.529–1.853; P<0.001; Figure 1D). Model calibration of the KAM score and observed data was acceptable (Brier score=0.073; goodness of fit P=0.748; Supplementary Figure 4).

Figure 1.

Korean Acute Myocarditis (KAM) score. Scoring system to predict the need for veno-arterial extracorporeal membrane oxygenation (VA-ECMO) in patients with acute myocarditis. (A) Variables required to calculate the KAM score. (B) Distribution of KAM scores in the training cohort. (C) Discriminant ability of the KAM score. (D) VA-ECMO implementation rate according to KAM score. CI, confidence interval; OR, odds ratio.

External validation of the predictive model (AUC 0.945; 95% CI 0.916–0.974) and KAM score (AUC 0.921; 95% CI 0.885–0.958) showed acceptable discriminant function (Figure 2).

Figure 2.

External validation of the predictive model and Korean Acute Myocarditis (KAM) score. Receiver operating characteristic curves and areas under the curve (AUC) for the (A) prediction model and (B) KAM score in the training and testing cohorts.

In-Hospital Mortality According to KAM Score and VA-ECMO

The in-hospital mortality rates of the training cohort, patients without VA-ECMO, and patients with VA-ECMO were 11.0%, 5.3%, and 33.3%, respectively (Supplementary Table 2). The KAM score was associated with the risk of in-hospital mortality in the total study population (OR 1.317; 95% CI 1.236–1.404; P<0.001), patients with VA-ECMO (OR 1.180; 95% CI 1.047–1.330; P=0.007), and in patients without VA-ECMO (OR 1.534; 95% CI 1.350–1.743; P<0.001; Figure 3A). When stratifying patients without VA-ECMO using a KAM score of 8, the cut-off value for predicting the implementation of VA-ECMO, patients with higher KAM scores exhibited more severe clinical characteristics than those with lower KAM scores (Supplementary Table 7). Consistent with these findings, as the KAM score increased beyond 8, there was a trend towards higher in-hospital mortality among patients without VA-ECMO than among those with VA-ECMO (Figure 3B).

Figure 3.

In-hospital mortality according to the Korean Acute Myocarditis (KAM) score and veno-arterial extracorporeal membrane oxygenation (VA-ECMO) implementation. (A) Odds ratios (with 95% confidence intervals) for KAM scores with respect to in-hospital mortality according to VA-ECMO implementation. (B) Event rate for in-hospital mortality according to VA-ECMO presented using spline curves.

Discussion

The present study identified factors associated with the implementation of VA-ECMO in patients with acute myocarditis and developed a simple scoring system capable of predicting the implementation of VA-ECMO (Figure 4). The main findings are as follows. First, initial mean blood pressure <65 mmHg, cardiac arrest, Glasgow Coma Scale score ≤12, platelet count <100×10³/mL, pulmonary congestion on chest X-ray, QRS interval ≥120 ms, left or right bundle branch block, and LVEF <40% were found to be significant predictors of the implementation of VA-ECMO in patients with acute myocarditis. Second, the KAM score, a new scoring system, developed from the final prediction model, showed excellent discriminant ability to predict the implementation of VA-ECMO. Third, the KAM score was also associated with the risk of in-hospital mortality. Furthermore, among patients with KAM scores above the cut-off value, in-hospital mortality tended to be higher among patients without VA-ECMO than among patients with VA-ECMO.

Figure 4.

Novel risk scoring model to assess the need for veno-arterial extracorporeal membrane oxygenation (VA-ECMO) in patients with acute myocarditis. AUC, area under the curve; CXR, chest X-ray; KAM score, Korean Acute Myocarditis score; LBBB, left bundle branch block; LVEF, left ventricular ejection fraction; RBBB, right bundle branch block.

Use of VA-ECMO in Patients With Acute Myocarditis

Considering the self-limiting nature of acute myocarditis, mechanical circulatory support, including VA-ECMO, can be used for bridge-to-recovery purposes.¹⁵ Conversely, in patients who cannot expect to recover from acute deterioration, VA-ECMO can be used as a bridge to transplantation or durable left ventricular assist device implantation. In this regard, VA-ECMO can be a useful tool to improve the prognosis of patients with acute myocarditis. Nevertheless, the question remains as to how to distinguish patients who may require VA-ECMO. This question is clinically relevant because some patients with acute myocarditis can deteriorate rapidly, and delaying the implementation of VA-ECMO can result in fatal outcomes.

However, because the prognosis and clinical course of myocarditis remain poorly understood, most previous studies have focused on comparing the prognosis according to clinical course (fulminant vs. non-fulminant).^5,12 It was only recently that these studies revealed that patients with fulminant myocarditis have a worse prognosis than those with non-fulminant myocarditis. Moreover, another study from the same group showed that, in addition to fulminant presentation, histologic subtype and QRS widening were associated with short- and long-term mortality.⁶ However, due to the insufficient number of patients included in that study, other predictors associated with prognosis could not be identified. Therefore, by analyzing data from a multicenter registry, which collected detailed data on initial clinical characteristics and included a sufficient number of patients, the present study aimed to identify predictors of VA-ECMO implementation. In this study, 8 key variables, comprising clinical and laboratory parameters at initial presentation, were associated with the implementation of VA-ECMO. The most notable finding of the present study is that the clinical course of patients with acute myocarditis can be predicted using simple clinical, electrocardiographic, and echocardiographic characteristics that can be easily assessed in clinical practice.

Scoring System to Predict VA-ECMO Implementation

In this study, we constructed a scoring system (KAM score) capable of quantifying the risk of VA-ECMO implementation and alerting physicians to prepare for VA-ECMO in the near future. To the best of our knowledge, this scoring system is the first model designed to predict the implementation of VA-ECMO in acute myocarditis. Furthermore, the discriminant ability of the KAM scoring system was excellent, and various validation processes assured the robustness of the system. Essentially, this scoring system was also associated with in-hospital mortality, regardless of hemodynamic status or the use of VA-ECMO. These findings imply that this scoring system is a clinically relevant and useful tool that can predict the clinical course of acute myocarditis. Nevertheless, caution is needed when using the KAM score in clinical practice. Indications for VA-ECMO in the present study were not standardized and detailed information on clinical situations leading to the implementation of VA-ECMO was collected only retrospectively. Accordingly, a positive KAM score should not be interpreted as an absolute indication for VA-ECMO and should be used to alert physicians that VA-ECMO may be needed depending on the patient’s clinical condition. If adopted in daily practice, this predictive tool may be used to stratify patients according to the expected risk and to implement appropriate management accordingly. Specifically, the use of the KAM score may prevent unnecessary examinations and transfer to high-level facilities among low-risk patients, while also enabling timely application of advanced management, including VA-ECMO, in high-risk patients, ultimately possibly improving the prognosis of acute myocarditis. Indeed, the fact that a substantial proportion of patients with high KAM scores and severe clinical characteristics died without VA-ECMO in the present study supports the clinical utility of the KAM score.

Study Limitations

Some limitations should be acknowledged. First, this study has innate limitations due to the retrospective design. Although all participating hospitals were tertiary care centers with expertise in managing acute myocarditis, patient management, including the use of VA-ECMO, was not standardized and was left to the discretion of the attending physician. The absence of a standardized protocol may have contributed to variability in treatment outcomes and potentially influenced the results of this study. Future prospective studies with standardized protocols, including clear indications for VA-ECMO, are warranted to address these limitations. Second, further external validation of the model in patient populations that received different management in different settings is needed. Although this study conducted external validation by dividing the participating centers into training and testing cohorts, participating centers shared similar patient characteristics. Given that all participating centers were tertiary hospitals and primarily enrolled patients with severe conditions, as evidenced by the high VA-ECMO implementation rate, further external validation is needed to assess the model’s generalizability to cohorts with a lower proportion of severe cases. Third, not all patients had a biopsy-proven diagnosis and patients with other diseases that mimic myocarditis may have been included in the cohort. However, the included patients all satisfied the diagnostic criteria suggested by the European Society of Cardiology position statement.⁸ In addition, the proportion of biopsy-proven myocarditis among the total patient population was higher or at least comparable to that reported in recent studies.^5,12

Conclusions

Using 8 variables, namely initial mean blood pressure, cardiac arrest, Glasgow Coma Scale score, platelet count, pulmonary congestion, QRS interval, left or right bundle branch block, and LVEF, the present study developed a novel risk scoring system predicting the implementation of VA-ECMO in patients with acute myocarditis. This KAM score could provide information regarding the clinical course of patients with acute myocarditis and could guide decisions to proceed with VA-ECMO.

Acknowledgment

The authors thank Joong Hyun Ahn (Biomedical Statistics Center, Institute of Future Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine) for support with statistical analysis.

Sources of Funding

This study was supported by a research grant from the Korean Cardiac Research Foundation (Grant no. 202201-01).

Disclosures

All authors declare that there is no conflict of interest relevant to the submitted work.

Author Contributions

D.H. contributed to study conception, design, and analysis, as well as drafting of the manuscript. J.H.Y. contributed to study conception, design, data acquisition, interpretation, and critical revision, and takes responsibility for the integrity of study. All other authors contributed to data acquisition, interpretation, and revision.

IRB Information

The study protocol was approved by the Institutional Review Board at Samsung Medical Center (Reference no. 2022-01-003).

Data Availability

The deidentified participant data will be shared on a request basis. Please contact the corresponding author directly to request data sharing. The entire dataset used will be available, including the study protocol. Data sharing will begin upon IRB approval at Samsung Medical Center and will continue indefinitely. Researchers will be granted access to the data provided they have a legitimate research purpose. Any analyses on the data will be approved and data will be shared as an Excel file via email.

Supplementary Files

Please find supplementary file(s);

https://doi.org/10.1253/circj.CJ-24-0684

References

• 1.   Basso C. Myocarditis. N Engl J Med 2022; 387: 1488–1500.
• 2.   Ammirati E, Moslehi JJ. Diagnosis and treatment of acute myocarditis: A review. JAMA 2023; 329: 1098–1113.
• 3.   Tschöpe C, Ammirati E, Bozkurt B, Caforio ALP, Cooper LT, Felix SB, et al. Myocarditis and inflammatory cardiomyopathy: Current evidence and future directions. Nat Rev Cardiol 2021; 18: 169–193.
• 4.   Kawano H, Ikeda S, Kanaoka K, Sato S, Eto R, Ueno Y, et al. Multisystemic inflammation influences prognosis in fulminant lymphocytic myocarditis. Circ J 2024; 88: 722–731.
• 5.   Ammirati E, Cipriani M, Moro C, Raineri C, Pini D, Sormani P, et al. Clinical presentation and outcome in a contemporary cohort of patients with acute myocarditis: Multicenter Lombardy Registry. Circulation 2018; 138: 1088–1099.
• 6.   Ammirati E, Veronese G, Brambatti M, Merlo M, Cipriani M, Potena L, et al. Fulminant versus acute nonfulminant myocarditis in patients with left ventricular systolic dysfunction. J Am Coll Cardiol 2019; 74: 299–311.
• 7.   Tara S, Yamamoto T, Kanaoka K, Oka E, Shiomura R, Matsuda J, et al. Effects of cardioprotective drugs on 90-day mortality or heart transplantation in patients with fulminant myocarditis. Circ Rep 2024; 6: 322–332.
• 8.   Caforio AL, Pankuweit S, Arbustini E, Basso C, Gimeno-Blanes J, Felix SB, et al. Current state of knowledge on aetiology, diagnosis, management, and therapy of myocarditis: A position statement of the European Society of Cardiology Working Group on Myocardial and Pericardial Diseases. Eur Heart J 2013; 34: 2636–2648, 2648a–2648d.
• 9.   Friedrich MG, Sechtem U, Schulz-Menger J, Holmvang G, Alakija P, Cooper LT, et al. Cardiovascular magnetic resonance in myocarditis: A JACC white paper. J Am Coll Cardiol 2009; 53: 1475–1487.
• 10.   Aretz HT. Myocarditis: The Dallas criteria. Hum Pathol 1987; 18: 619–624.
• 11.   Diddle JW, Almodovar MC, Rajagopal SK, Rycus PT, Thiagarajan RR. Extracorporeal membrane oxygenation for the support of adults with acute myocarditis. Crit Care Med 2015; 43: 1016–1025.
• 12.   Ammirati E, Cipriani M, Lilliu M, Sormani P, Varrenti M, Raineri C, et al. Survival and left ventricular function changes in fulminant versus nonfulminant acute myocarditis. Circulation 2017; 136: 529–545.
• 13.   Lorusso R, Centofanti P, Gelsomino S, Barili F, Di Mauro M, Orlando P, et al. Venoarterial extracorporeal membrane oxygenation for acute fulminant myocarditis in adult patients: A 5-year multi-institutional experience. Ann Thorac Surg 2016; 101: 919–926.
• 14.   Nagai T, Inomata T, Kohno T, Sato T, Tada A, Kubo T, et al. JCS 2023 guideline on the diagnosis and treatment of myocarditis. Circ J 2023; 87: 674–754.
• 15.   Nishimura T, Hirata Y, Ise T, Iwano H, Izutani H, Kinugawa K, et al. JCS/JSCVS/JCC/CVIT 2023 guideline focused update on indication and operation of PCPS/ECMO/IMPELLA. Circ J 2024; 88: 1010–1046.
• 16.   Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE Jr, Drazner MH, et al. 2013 ACCF/AHA guideline for the management of heart failure: A report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013; 62: e147–e239.