Patient selection and data collection
We retrospectively analysed our institution’s ECMO database to identify patients requiring circulatory support with ECMO for suspected or confirmed PE between January 2012 and December 2019. According to the American Heart Association and European Society of Cardiology guidelines, PE was diagnosed using the diagnostic strategy tools (3, 4). Patients presenting with hemodynamic instability were defined as massive PE, and imaging and laboratory parameters were used to distinguish submassive and low-risk PE (3). Baseline characteristics and clinical variables were collected from patients’ medical records, including age, sex, body mass index, smoking history, comorbidities, predisposing factors for venous thromboembolism, and presenting symptoms and signs. The time of presence of initial symptoms, refractory circulatory collapse (cardiogenic shock or cardiac arrest), and ECMO initiation was recorded. The prognostic values for PE were collected, including arterial blood gas data, cardiac troponin I (9), electrocardiogram (ECG), echocardiography, and computed tomography (CT). Cardiac troponin I level ≥ 0.04 ng/mL (10% the coefficient of variation levels) was defined as elevated cardiac troponin I level in our institution. ECG findings of RV strain (10, 11) were defined as by the presence of at least one of the following: 1) S1Q3T3 pattern, 2) complete or incomplete RBBB, 3) inverted T waves in leads V1–V4, 4) ST elevation in lead aVR, 5) qR pattern in lead V1. Echocardiographic findings of RV dysfunction (12, 13) before or after ECMO initiation were defined as by the presence of at least one of the following: 1) RV dilatation or dysfunction, 2) tricuspid annular plane systolic excursion < 16 mm, 3) D-shaped left ventricle, 3) dilated inferior vena cava without respiratory variation. At CT scan, before or after ECMO initiation, RV dysfunction (14) was defined as a right-to-left ventricular (RV/LV) diameter ratio ≥ 0.9 measured in the transverse or four-chamber view.
Indication, technique, and weaning of ECMO
All patients with persistent profound hypotension or cardiac arrest were placed on ECMO by cardiovascular surgeons. Cannulation was performed under ultrasound guidance or surgical exploration, with a 21-French to 23-French venous cannula inserted in the femoral vein, and a 15-French to 16.5-French arterial cannula inserted in the femoral artery. A 6-French distal perfusion catheter was placed in the ipsilateral superficial femoral artery for patients with significant leg ischemia. Anticoagulant therapy with unfractionated heparin was mandatory unless contraindicated, with a target activated clotting time of 180 to 220 s. Intravenous antihypertensive agents were administered to maintain systolic blood pressure below 120 mmHg to prevent brain hemorrhage, and inotropes or vasopressors were used to achieve target mean arterial pressure of 60 to 80 mmHg. Once the patient had markedly recovered after definitive treatment, such as anticoagulation therapy, or thrombolysis, ECMO was weaned by gradually reducing circuit flow till 1 liter per minute. The cannulas were removed if the patient could maintain hemodynamic stability and acceptable pulmonary artery pressure while the flow was temporarily interrupted. If the patient had a brain death, ECMO support was withdrawn early.
Outcomes variables
The primary outcome was in-hospital mortality. Secondary outcomes included acute neurologic complications (profound comatose states, seizures, brain hemorrhage, brain infarction, and brain death), acute kidney injury that required renal replacement therapy, major bleeding that required endoscopic, endovascular or surgical intervention, and ECMO-related complications, including vascular access site complications and leg ischemia.
Statistical analysis
Continuous variables were expressed as mean ± standard deviation and were analysed using Student's T test or Mann-whitney U-test. Categorical variables were expressed as number (percentage) and were compared using the Chi-squared test. We used RV/LV diameter ratio measurement on CT and cardiac troponin I prior to refractory circulatory collapse for predicting the clinical course. Correlations between the CT values or the cardiac biomarker and the time interval from diagnosis to persistent profound hypotension or cardiac arrest were evaluated using Pearson’s correlation or Spearman correlation. P values less than 0.05 were considered to indicate significant differences. All analyses were conducted using SPSS software (version 19.0; SPSS Inc, Chicago, IL).