Outcome of Prolonged Cardiac Arrest Under Extracorporeal Cardiopulmonary Resuscitation Due to Acute Myocardial Infarction: Complete vs Culprit Revascularization

Immediate revascularization of infarct-related vessel (culprit strategy) is effective in patients with acute myocardial infarction (AMI) with cardiogenic shock. However, for AMI patients complicated with prolonged cardiac arrest under extracorporeal membrane oxygenation resuscitation (ECPR), whether culprit revascularization (IR) or complete revascularization (CR) is associated with better clinical outcome is not known. Methods: Patients with AMI complicated with prolonged cardiac arrest under ECMO support between 2006 and 2016 were included and were grouped by the status of revascularization completeness in three coronary territories into IR and CR groups. The primary endpoint is favorable neurological outcomes at hospital discharge. The second endpoint is the probability of a composite of major adverse cardiac and cerebral events (MACCE) at 1-year follow-up. AMI: acute myocardial infarct, CCI: Charlson comorbidity index, CK-MB: creatine kinase MB iso-enzyme, CKD: chronic kidney disease, COPD: chronic obstructive pulmonary disease, CPR: cardiopulmonary resuscitation, PR: partial revascularization, CVA: cerebral vascular attack, DM: diabetes mellitus, ECMO: extracorporeal membrane oxygenation, IABP: intra-aortic balloon pumping, NYHA: New-York Heart association, OHCA: out-of-hospital cardiac arrest, PAOD: peripheral artery occlusive disease TIA: transient ischemic attack, CR: complete revascularization, UR: unsuccessful revascularization, VF: ventricular brillation, VT: ventricular tachycardia. Initial CAD ischemic regions are counted by the number of involvements of left anterior descending (LAD), left circumex (LCX), right coronary artery (RCA), with left main (LM) is viewed as involvement of LAD and LCX. PCI: percutaneous coronary intervention, including balloon angioplasty or stenting. Low-ow duration: duration between CPR and ECMO run.


Introduction
Acute myocardial infarct (AMI) complicated with cardiogenic shock and prolonged cardiopulmonary resuscitation (CPR) needing extracorporeal membrane oxygenation (ECMO) support is one of the most detrimental conditions for patients and the most challenging situations for medical personnel. Because rapid death due to persistent shock is likely in this condition, several clinical trials, such as CHEER trial (1) and Prague Out-of-Hospital Cardiac Arrest (OHCA) study, (2) have advocated immediate revascularization of occluded coronary arteries. However, these studies have not compared the probabilities of hospital survival and favorable neurological outcomes associated with partial revascularization (PR) or complete revascularization (CR). Theoretically, CR of all stenotic coronary arteries, either through percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG), may provide the highest chance of heart function recovery, but it may also induce end-organ damage because of the prolonged procedural time and repeated hypoperfusion of the brain and other vital organs during revascularization. The CULPRIT-SHOCK study demonstrated a similar survival rates between PR and CR patients but with a higher incidence of repeated coronary intervention in PR patients during 1-year follow-up.
In the present study, the effects of revascularization strategy on hospital survival and 1-year outcomes for an extremely critical AMI patients, namely patients with cardiac arrest and requiring ECPR, were evaluated. The results of the study can contribute to decision-making protocols(3) for the detrimental AMI complicated by profound cardiogenic shock.

Methods
All adult patients (age ≥18 years) who had received ECPR and emergency coronary angiography at our institution between 2006 and 2016 were included. All demographic and clinical data were prospectively collected and systemically reviewed under institutional approval (NTUH-9561707007.) The present study was approved by the review board of our institute (NTUH 201810079 RIN).
The contraindications of ECPR, modi ed from previous reports, (4,5) are listed in supplement Table 1. The decision to perform coronary angiography after ECPR was made by the consulted cardiologist based on a combination of multiple factors, including presenting symptoms before cardiac arrest, risk factors for coronary artery disease (CAD), and initial and serial electrocardiography (EKG) ndings suspecting AMI or a VT or VF rhythm. Emergency coronary angiography was performed on emergency status after exclusion of acute aortic dissection or massive intracranial hemorrhage detected by computed tomography scanning. After con rming the diagnosis of CAD, the decision of emergency PCI or CABG surgery was determined by a discussion between the on-duty cardiologist and cardiac surgeon (10 cardiologists and 5 cardiac surgeons on call 24 h a day and 7 days a week) in which multiple factors, including anatomy of the coronary arteries, culprit coronary lesion pathology, general conditions before the index ECPR event, low-ow duration before ECMO, and the patients' and relatives' will, were considered.  AMI: acute myocardial infarct, CCI: Charlson comorbidity index, CK-MB: creatine kinase MB isoenzyme, CKD: chronic kidney disease, COPD: chronic obstructive pulmonary disease, CPR:  cardiopulmonary resuscitation, PR: partial revascularization, CVA: cerebral vascular attack, DM:  diabetes mellitus, ECMO: extracorporeal membrane oxygenation, IABP: intra-aortic  Low-ow duration: duration between CPR and ECMO run.
Emergency PCI was performed using the femoral artery approach. The coronary lesions were rst prepared through balloon angioplasty followed by stent implantation. The activating clotting time was maintained in the range of 250-300 s through continuous heparin infusion; intracoronary IIb/IIIa inhibitors or aspiration catheter were administered if the thrombus burden was large or a TIMI 3 ow was not obtained. Those patients with post-PCI TIMI 1 or 2 ow or with coronary anatomy inappropriate for PCI, emergency CABG was performed soon with ECMO support. Side-clamping of the ascending aorta was applied for saphenous vein graft anastomosis to the ascending aorta in most cases. Anastomosis of target coronary arteries was performed under a beating heart condition in most patients. Induced cardiac arrest with cardioplegia was used in patients with di cult exposure of target vessels or very hemorrhagic and fragile myocardium.

Exclusion criteria
The patients who experienced cardiac arrest and received CPR < 10 min before the return of spontaneous circulation (ROSC) were excluded from the study because they did not ful l the criteria of prolonged CPR. Patients with acute aortic dissection, massive intracranial hemorrhage, prosthetic valve dysfunction, endocrine crisis, or failure of ROSC after ECMO support were excluded from further coronary intervention.
Patients who received CPR and ECMO after coronary intervention or CABG during the same admission were excluded. Transferred patients who did not receive their index ECPR in the study hospital were also excluded because of unstandardized de nitions of arrest time, CPR duration, and ECMO management protocol. Patients who experienced shock necessitating ECMO in an elective or emergency situation without cardiac massage were also excluded from the study. Patients with clinical presentation of malignant arrhythmia with patent coronary arteries were excluded. Patients refused to receive PCI or CABG after coronary angiography study were also excluded from the study. Patients with incomplete data of CPR, follow-up, and coronary angiography were also excluded.

ECMO setup
ECMO equipment and management has been described elsewhere. (5,6) The principal components of the ECMO circuit used on our patients were centrifugal pumps and an oxygenator (Medtronic, Anaheim, CA, USA; Medos, Stolberg, Germany; Maquet, Rastatt, Germany). The circuit was pre-assembled without priming and was then primed using saline containing 2 U/mL heparin on receiving an ECPR call.
Review of coronary angiography data to determine revascularization types was retrospectively performed by two independent cardiologists (L.Y.L. and W.Y.C.). Coronary artery disease was de ned as coronary vessels with stenosis > 70% or left main vessel > 50% stenosis by simple visual estimation. Complete revascularization (CR) was de ned by each of the involved territories (left anterior descending [LAD], left circum ex [LCX], and right coronary artery) being successfully revascularized through PCI or CABG 24 hours after the index CPR episode. Incomplete revascularization (IR) was de ned as culprit vessel only intervention or culprit vessel plus another artery intervention, leaving at least one diseased coronary artery territory. Successful coronary angioplasty, through either balloon angioplasty or coronary stenting, was de ned as a TIMI 3 ow at the end of the emergency procedure. When left main coronary artery is involved, the patient is considered to have a two-vessel disease (LAD or LCX.)

Outcomes
The primary endpoint of the study was a favorable neurological outcome (Cerebral Performance Category [CPC] 1 or 2) at hospital discharge. The second endpoint is the composite of major adverse cardiac and cerebral events (MACCE) de ned as cardiac death, myocardial infarction (MI), revascularization, or stroke at 1 year since the index ECPR. Successful weaning of ECMO is de ned as separation from ECMO support without subsequent mortality in the next 48 h. Safety endpoints were adverse events of infection, limb ischemia, renal failure, bleeding, or stroke during hospitalization among different study groups. The follow-up data of the patients who survived ECPR to hospital discharge were collected through chart review.

Statistical analysis
All demographic data and post-ECMO events data were prospectively collected from an ECMO registration database, from which several studies regarding ECPR had been reported.(5) − (7) (8) Continuous variables are presented as mean ± standard deviation and were examined using the Student t test and analysis of variance (ANOVA). Categorical variables were analyzed using the chi-square test or Fisher exact test, depending on sample sizes. The variables affecting the major outcomes (i.e., hospital survival and favorable neurological function) that were signi cant (P < 0.15) in the univariate analysis were included in multivariate logistic regression with backward stepwise analysis. The patients' 1-year survival was modeled using the Kaplan-Meier method and tested using the log-rank test and Cox regression. Statistical signi cance was set at P < 0.05 and statistical signi cance was set at 0.10 > P > 0.05. All statistical analyses were performed using MedCalc (version 17.8.6; MedCalc Software, Ostend, Belgium). All statistical analyses were reviewed by a specialist (J.W.L) blinded to the present study design.

Results
A total of 502 adult ECPR patients during the study period was retrieved from a prospectively collected database and retrospectively analyzed. Among the patients, 115 patients were suspected with acute coronary syndrome and underwent emergency coronary angiography study. After excluding the patients who did not receive revascularization procedure (13), with patent coronary arteries (7), or with incomplete angiography data (5), 90 patients were included in the present study (Fig. 1) Fig. 1).
The demographic data of both study groups are presented in Table 1. In brief, most demographic data are comparable between IR and CR groups except the extent of coronary artery involvement. The number of ischemic territories was higher in IR group (2.7 ± 0.5) than that in CR group (1.9 ± 0.9) (P,0.001).
The safety endpoints among the different study groups are listed in Table 2. The incidence of infection, limb ischemia needing fasciotomy or amputation, renal failure needing mechanical renal replacement therapy, brain anoxia, stroke (ischemic or hemorrhagic), and bleeding are all comparable among the three groups. This nding indicates that CR group, even with more CABG procedure than IR group, is not associated with increased risk of peri-procedural morbidity. It is noteworthy that ischemic stroke rate is higher in CR group (12.1%) as compared with IR group (3.1%), even though which does not achieve statistical signi cance (P = 0.156.) * denotes a trend of statistical signi cance. ** denotes statistical signi cance.
Hospital outcomes are also summarized in Table 2. In brief, ECMO days, ventilator days, days of intensive care unit, and hospitalization days were comparable among both groups. One patient received ventricular assist device (VAD) support and another one received heart transplantation (HTX). The rate of successful wean-off of ECMO without VAD or HTX was higher in the CR (60.3%) than that in IR (43.7%) but not attain statistical signi cance (P = 0.132). Hospital survival is also marginally higher in CR (48.3%) than that in IR (28.1%) (P = 0.064). Favorable neurological outcome at hospital discharge (CPC 1, 2) was also higher in CR group (37.9%) than in IR group (21.9%), but it does not achieve statistic signi cance (P = 0.121) ( Table 2).
Logistic regression analysis of favorable neurological outcome at hospital discharge is presented in Table 3. The univariate logistic regression analysis revealed that only age is a potentially signi cant factor (odds ratio [OR] for every 10-year increase in age is 0.66, P = 0.078). The low-ow duration, CABG, or CR are not shown as signi cant risk factors. Multivariate logistic regression is not performed because only one factor (age) is found by univariate logistic regression analysis. In 1-year follow-up of the hospital survivors, mortality was recorded in four cases, all attributable to cardiac causes (one in IR and three in CR groups.) Morbidity was recorded as following: two patients received PCI (one each in IR and CR groups), one received CABG (in the IR group), and one had stroke (in IR group). In total, six cases with MACCE in 1-year follow-up period (three in IR and three in CR groups.) Kaplan-Meier survival analysis for 1-year freedom from MACCE probability for both study groups are illustrated in Fig. 2, revealing marginally higher probability in CR group than in IR groups (42.4% for CR; 24.6% for IR, P = 0.051). Taking only hospital survivors into consideration, the freedom from MACCE probability at 1-year is comparable between CR group (87.5%) and IR group (67.5%) (P = 0.162.) Cox regression analysis for 1-year MACCE probability for all study patients is presented in Table 4. Univariate analysis revealed that age, Euroscore II, low-ow duration, initial shockable rhythm, CAD number, and CR were potential signi cant factors. In multivariate analysis, only age, low-ow duration, initial shockable rhythm, and CR were revealed as signi cant factors (hazard ratios: 1.37 for every 10-year increase in age, 1.23 every 10-minute increase in low-ow duration, 0.53 for initial shockable rhythm, and 0.53 for CR; P = 0.002, 0.027, 0.027, and 0.020, respectively).

Discussion
The present study is the largest series focusing on the impact of revascularization strategy for AMI patients complicated with cardiac arrest under ECMO support. The results indicated that CR patients is associated with insigni cantly higher rate of favorable neurological outcomes at hospital discharge, and lower MACCE probabilities at 1-year follow-up than IR patients. The results might contribute to the decision-making in revascularization strategy in this extremely critical condition.
AMI complicated with cardiogenic shock is a highly emergent condition needing immediate decision to support the critical hemodynamics and protect the jeopardized myocardium.(9) , (10) Data from IABP-SHOCK II trial revealed a 30-day mortality of 43.6% in patients with AMI complicated by cardiogenic shock with or without IABP support. (11) Several studies have advocated ECMO use during primary PCI for AMI with cardiogenic shock.(12) , (13) Studies have also advocated primary PCI for patients with AMI after ECMO use to improve hospital survival rates. (14) The mortality reported in the present study was considerably higher than that reported in the SHOCK I or IABP-SHOCK II trials because the present study group comprised a relatively severe patients in dying condition. However, the present study outcomes were comparable with those of our previous studies on ECPR.(4) − (6) , (7) Previous study (15) had reported a systemic approach to OHCA patients, including ECMO initiation and immediate coronary angiography in patients with presumed ischemic etiology increased the rate of favorable neurological outcomes at hospital discharge from 3.3% in 2009 to 8.5% in 2013. However, the types of coronary revascularization in that study was not reported in detail. Other studies have suggested infarct-related PCI (e.g., HORIZONS-AMI study,(16) APEX-AMI trial (17)) in patients with AMI without cardiogenic shock, whereas others have suggested multivessel PCI (e.g., SWISSI II trial (18) and HELP-AMI (19)). The CULPRIT-SHOCK trial demonstrated that in patients with AMI complicated by cardiogenic shock, immediate culprit-lesion-only PCI had better clinical outcomes than immediate multivessel PCI in patients with AMI and cardiogenic shock. (20) , (21) However, the CULPRIT-SHOCK trial did not include patients with prolonged resuscitation. (20) The results of a 1-year follow-up study of the CULPRIT-SHOCK trial (21) showed that although the mortality rates did not differ signi cantly between the culprit-lesiononly and multivessel PCI groups, the incidence of rehospitalization and repeated intervention was considerably higher in the culprit-lesion-only PCI group than in the multivessel PCI group during the follow-up. This nding indicates that for patients with AMI without intractable cardiogenic shock, the bene t of solving all residual myocardial ischemic burdens does not outweigh the risk of the procedure, but it needs to be treated during the follow-up period. Theoretically, the presence of silent myocardial ischemic burden is associated with a negative effect on long-term outcome. [21][22][23] The necessity to treat residual myocardial ischemic burden is supported by other meta-analysis, (22) in which during primary PCI in patients with AMI without cardiogenic shock, CR was associated with a signi cantly lower total mortality and MI risk during follow-up than PR was. In addition, a post hoc analysis of the IABP-SHOCK II trial (23) indicated that the presence of chronic total occlusion in a non-infarction-related vessel is associated with an increased ventricular arrhythmia incidence and mortality at 12-month follow-up after index AMI. However, whether non-infarct-related myocardial ischemic burden needs to be immediately treated in detrimental AMI complicated with cardiac arrest needing ECMO support is unanswered in those studies. From the results of the present study, considering that CR did not increase the early mortality in this group, we suggest CR for the patients to achieve the best chance of hospital survival with favorable neurological status and high 1-year freedom from MACCE probability.
Regarding the procedural risk and complications, data of the present study indicate that CR group, even with more CABG procedure than IR group, is not associated with increased risk of peri-procedural morbidity. It is noteworthy that ischemic stroke rate is higher in CR group (12.1%) as compared with IR group (3.1%), even though which does not achieve statistical signi cance (P = 0.156.) Considering that emergency more CABG was performed in CR group with ECMO support as compared with IR group, and using saphenous vein grafts in most cases, the high incidence of ischemic stroke might be explained by inadequate cardiac support by ECMO during exposure of the posterior ventricle in the CABG operation, and embolization during side-clamping of the ascending aorta during anastomosis of the proximal vein grafts. In addition, early studies indicates that the incidence of postoperative pneumonia was also higher in CABG patients as compared with other types of operation. (24,25) This high incidence of post-CABG pneumonia can be explained by prolonged pump time, excessive blood transfusion, prolonged ICU stay, and limited motility due to IABP and ECMO use, all of which are associated with risk factors for postoperative pneumonia. (24) Low-ow duration has generally been recognized as a signi cant risk factor for neurological outcome (7)  EKG was not obtainable for all patients during the ECPR condition to clearly de ne ST-elevation MI (STEMI) or non-STEMI. However, according to their severity and emergency, the treatment protocol was based on the guideline for STEMI.
The present study was not a randomized trial. However, the decision of revascularization strategy was made by a joint discussion between on-duty cardiologists and cardiac surgeons. It may be considered as real-world data re ecting the true clinical scenario.

Conclusion
For AMI patients complicated with cardiac arrest supported with ECMO, CR, as compared with IR, had similar outcome in one-year MACCE outcomes. Therefore, we suggest culprit strategy may be acceptable for these severe AMI patients under ECMO support in the point of saving procedure duration.