Simultaneous Venoarterial Extracorporeal Membrane Oxygenation and Percutaneous Left Ventricular Decompression Therapy with Impella Is Associated with Improved Outcomes in Refractory Cardiogenic Shock

Abstract

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) has been used for refractory cardiogenic shock; however, it is associated with increased left ventricular afterload. Outcomes associated with the combination of a percutaneous left ventricular assist device (Impella) and VA-ECMO remains largely unknown. We retrospectively reviewed patients treated for refractory cardiogenic shock with VA-ECMO (2014–2016). The primary outcome was all-cause mortality within 30 days of VA-ECMO implantation. Secondary outcomes included duration of support, stroke, major bleeding, hemolysis, inotropic score, and cardiac recovery. Outcomes were compared between the VA-ECMO cohort and VA-ECMO + Impella (ECPELLA cohort). Sixty-six patients were identified: 36 VA-ECMO and 30 ECPELLA. Fifty-eight percent of VA-ECMO patients (n = 21) had surgical venting, as compared to 100% of the ECPELLA cohort (n = 30) which had Impella (±surgical vent). Both cohorts demonstrated relatively similar baseline characteristics except for higher incidence of ST-elevation myocardial infarction (STEMI) and percutaneous coronary intervention (PCI) in the ECPELLA cohort. Thirty-day all-cause mortality was significantly lower in the ECPELLA cohort (57% vs. 78%; hazard ratio [HR] 0.51 [0.28–0.94], log rank p = 0.02), and this difference remained intact after correcting for STEMI and PCI. No difference between secondary outcomes was observed, except for the inotrope score which was greater in VA-ECMO group by day 2 (11 vs. 0; p = 0.001). In the largest US-based retrospective study, the addition of Impella to VA-ECMO for patients with refractory cardiogenic shock was associated with lower all-cause 30 day mortality, lower inotrope use, and comparable safety profiles as compared with VA-ECMO alone.

Cardiogenic shock (CS) represents one of the most lethal conditions in modern-day medicine with mortality rates greater than 50%.¹ With approximately 50,000 cases annually in the United States, the need to establish more effective therapies is paramount.² Percutaneous hemodynamic support devices such as Impella have become an integral part of therapy for CS, in an effort to unload the left ventricle (LV), increase effective cardiac output, and thereby reverse metabolic derangements by improving end-organ perfusion without increasing myocardial oxygen demand.³ The level of support provided by the Impella platform (2.5–5.0 L/min) alone may not be sufficient in a subset of patients, especially those with cardiac arrest without return of spontaneous circulation, biventricular failure, profound metabolic derangements, and impaired oxygenation.

Peripheral venoarterial extracorporeal membrane oxygenation (VA-ECMO) provides a mechanism for complete hemodynamic and respiratory bypass in CS. Prior studies have shown that this type of therapy may confer a modest mortality benefit in CS.^4–6 Further, with the insertion of peripheral inflow venous and outflow arterial cannulae, there is retrograde injection of blood into the aorta and a subsequent increase in LV afterload, which has the potential for cavity distension, ventricular blood pool stasis, progressive myocardial ischemia, and persistent pulmonary edema with irreversible lung damage.^3,4,7 Consequently, these changes may result in coronary, ventricular, and pulmonary arterial thrombosis further increasing mortality.^8–11 Finally, the use of VA-ECMO alone is potentially detrimental to long-term myocardial recovery and may result in a more frequent need for durable LV assist device (LVAD) and cardiac transplantation.^12–14

Several studies have demonstrated that percutaneous^15,16 or surgical (LV, pulmonary arterial, or left atrial) venting may mitigate the ECMO-related concerns related to increased afterload and ventricular dilation, but require more invasive procedures with greater risk.^{8–11,17–21} With the advent of percutaneous microaxial flow pumps in the Impella platform, retrograde, transaortic unloading of the LV is now possible and provides a similar mechanism of decompression with less invasiveness.²² However, whether the addition of Impella to VA-ECMO improves outcomes in patients with refractory CS is still unclear. We therefore sought to review our experience with VA-ECMO in refractory CS with and without the use of concomitant Impella support.

Methods

Study Population and Definitions

We retrospectively reviewed all consecutive patients with a diagnosis of refractory CS treated with VA-ECMO at University Hospitals Cleveland Medical Center between 2014 and 2016. All patients underwent assessment by a multidisciplinary team of interventional cardiologists, intensivists, heart failure specialists, and cardiothoracic surgeons: Shock Team. The Shock Team used a combination of abnormal hemodynamics (systolic blood pressure < 90 mm Hg, cardiac index (CI) < 1.8 L/min/m² without hemodynamic support or <2.2 L/min/m² with support, pulmonary capillary wedge pressure >18 mm Hg), metabolic derangements (pH < 7.4, lactate > 4 mmol/L), and clinical findings consistent with systemic hypoperfusion despite adequate intravascular volume resuscitation and the use of inotropic and vasoactive agents to determine if refractory CS was present. Inotropes, vasopressors, and the initiation of VA-ECMO or Impella device insertion was determined by the Shock Team. In a majority of cases, Impella was inserted concomitantly or within 24 hours of VA-ECMO initiation (ECPELLA cohort).

Patients’ characteristics including demographics, laboratory parameters, hemodynamic recordings, vasoactive medications, clinical course, all-cause mortality, and associated complications were obtained from the electronic medical record. A Survival After VA-ECMO (SAVE) score based on diagnosis, age, weight, cardiac, respiratory and renal function was calculated for each patient and provides a validated method for predicting survival in CS patients treated with VA-ECMO.²³ A maximum inotropic score was also calculated for the first 3 days after the initiation of mechanical circulatory support (CS), which helps to convert the dosages of multiple different vasopressors and inotropes into a uniform numerical score to facilitate comparison.²⁴ Bleeding academic research consortium (BARC) definitions were used to categorize bleeding events.²⁵ Accepted consensus standards were used to define renal dysfunction and stroke endpoints.²⁶ Hemolysis was defined as elevation in lactate dehydrogenase (LDH) greater than 1,000 U/L associated with appropriate elevations in schistocytes, reticulocytes, or plasma-free hemoglobin in at least two consecutive blood samples. The primary outcomes were 30 day (and 1 year) all-cause mortality. Secondary outcomes were time on VA-ECMO support, inotropic score, need for dialysis, stroke, clinical hemolysis, and major bleeding events. The institutional review board at University Hospitals Cleveland Medical Center approved this study.

Management of Mechanical Support

The VA-ECMO implantation was performed with the assistance of dedicated perfusionists, and cardiothoracic surgeons/interventional cardiologists, depending on the clinical circumstance, at the bedside, in the cardiac catheterization laboratory, or operating room. If insertion was possible via a percutaneous approach, standard Seldinger access of the femoral artery and vein was attempted; femoral artery angiography was performed to ensure appropriate vessel caliber and anatomy (tortuosity, calcification, aneurysms, etc.) when fluoroscopy was available. When percutaneous cannulation was not feasible, surgical access and cannulation was performed at the discretion of the cardiothoracic surgeon in the operating room. The axillary or femoral artery was preferred arterial access points. A Dacron graft (8 mm diameter) was sewn end-to-side onto the axillary artery if it was the selected route. Standard-sized 15–19 French arterial cannulae and 23–25 French multihole venous cannulae were inserted. The VA-ECMO circuit consisted of venous outflow and arterial inflow cannulae attached to an external centrifugal pump and an oxygenator. The addition of an intraaortic balloon pump (IABP) or surgical vents (LV, pulmonary artery, left atrial) was left to the discretion of the managing cardiothoracic surgeon. Intravenous heparin infusion was initiated once access was secured to maintain an activated clotting time of 150–200 seconds to prevent thrombosis of the circuit. The flow parameters of the circuit were left to the discretion of the Shock Team, whose primary goal was to reverse the circulatory and metabolic derangements as measured by clinical and hemodynamic parameters.

The Impella device, described in detail elsewhere,²⁷ is a microaxial, nonpulsatile flow pump that is currently available in three sizes: Impella 2.5, Impella CP, and Impella 5.0. The choice of Impella implantation, which was performed either percutaneously or surgically, was based on a multitude of factors which included the level of support needed, feasibility of arterial access, and the need for concomitant percutaneous intervention or cardiac surgery. The Impella 2.5 and CP devices were typically implanted under fluoroscopic guidance after confirmation of appropriate femoral arterial caliber and anatomy by an interventional cardiologist. The Impella 5.0 device was inserted via the axillary artery through a Dacron graft interposition by a cardiothoracic surgeon, also under fluoroscopic guidance. All devices were initially set to max speed of flow (level P8) then lowered to level P3–P6 to avoid significant hemolysis during support. Impella position was assessed by echocardiography or fluoroscopy (if available) and P level was adjusted accordingly. The purge solution through the Impella system was dextrose-based fluid with or without heparin at a rate of 4–12 ml/hr to prevent pump thrombosis. Activated clotting times were to be maintained at 160–180 seconds per manufacturer recommendations while the pump remained on and functioning. Patients were monitored for clinical hemolysis, suction events, catheter thrombosis, and malpositioning and managed appropriately to ensure optimal flow.

Explantation of Devices

The VA-ECMO device was typically weaned at the discretion of the Shock Team with a gradual reduction in the circuit flow as directed by hemodynamic and clinical evidence of improving myocardial function. The primary goal was to minimize time of VA-ECMO support. If decannulation was indicated, the patient was brought to the operating room and the arterial and venous access sites were primarily repaired. Impella 2.5, CP, and 5.0 explantation was performed when appropriate hemodynamic and metabolic parameters were achieved and maintained with the lowest level of support (P2) for greater than 4–6 hours. If the 2.5 or CP device was used, standard explantation techniques including manual compression (and if necessary surgical open repair) were employed depending on the circumstance based on manufacturer recommendations. If an Impella 5.0 device was to be explanted, this was done in the operating room with suture occlusion of the Dacron graft on the axillary artery.

Statistical Analyses

Categorical variables are presented as numbers and percentages, and compared using χ² test. Continuous variables are presented as median and quartiles, and compared using Mann–Whitney U test. Survival analyses were performed with the Kaplan–Meier method and compared using log rank (Mantel-Cox) test. Multivariable-adjusted survival analyses were done using Cox proportional hazard models, adjusting for baseline characteristics that were statistically different between the cohorts. Interaction between characteristics and mortality among the cohorts were also performed using Cox proportional hazard models. All tests are two sided and p < 0.05 was considered statistically significant. All analyses were done using Statistical Package for Social Studies (SPSS, version 21).

Results

Baseline Characteristics

A total of 66 patients were treated for refractory CS using VA-ECMO for circulatory support between 2014 and 2016: 30 patients with ECPELLA and 36 patients with VA-ECMO. Baseline demographic characteristics were similar between the two cohorts except for ST-elevation myocardial infarction (STEMI) at presentation (ECPELLA: 50%, VA-ECMO: 17%, p = 0.007) and use of percutaneous coronary intervention (PCI) (ECPELLA: 30%, VA-ECMO: 2.8%, p = 0.004) as a part of the initial management strategy. There were no differences in physiologic and laboratory parameters between the two cohorts (Table 1). Cardiopulmonary resuscitation (CPR) was performed in 42% of the VA-ECMO and 40% of the ECPELLA groups (p = 1.0). Additional forms of surgical venting or IABP were used in various patients in both cohorts. Approximately 31% (n = 11) of the VA-ECMO and 20% (n = 6) of the ECPELLA cohort had some form of surgical vent placed. Intraaortic balloon pump was used in 28% (n = 10) and 17% (n = 5) in the VA-ECMO and ECPELLA cohorts, respectively. No venting or IABP use was observed in 14% (n = 5) of the VA-ECMO cohort. Simultaneous surgical venting and IABP use was observed in 28% (n = 10) of the VA-ECMO group and 10% (n = 3) of the ECPELLA group. In the ECPELLA cohort, 24 received Impella CP (80%), four received Impella 5.0 (13.3%), and two used Impella 2.5 (6.7%). Median time from diagnosis to VA-ECMO insertion was similar in both groups (ECPELLA: 14 hours [6–32] vs. VA-ECMO: 14 hours [3–24]; p = 0.38). Median time from VA-ECMO to the addition of Impella support was 14 hours [4,33]. Six patients had Impella before VA-ECMO (20%), 20 patients had VA-ECMO and Impella placed in the same setting (66.7%), and 4 patients had Impella placed after VA-ECMO (13.3%).

Table 1.: Baseline Characteristics

Primary Outcomes

At 30 days after VA-ECMO implantation, 45 patients died with a cumulative mortality rate of 68%. Mortality was significantly lower in the ECPELLA cohort compared with VA-ECMO (57% vs. 78%, hazard ratio [HR] 0.51 [0.28–0.94], log rank p = 0.02). This difference remained significant after adjusting for STEMI and PCI (adjusted HR 0.40 [0.19–0.84]; p = 0.016). One-year all-cause mortality for the entire cohort was 79% (n = 46). This was also significantly lower in the ECPELLA group versus VA-ECMO (ECPELLA: 69%, VA-ECMO: 87%; HR 0.52 [0.29–0.93], log rank p = 0.02) (Figure 1A, B), which remained statistically different after adjusting for STEMI and PCI (adjusted HR 0.39 [0.19–0.81]; p = 0.011). Of those patients who had CPR (ECPELLA: n = 12; VA-ECMO: n = 15), mortality rates were not statistically different between the two groups (ECPELLA: 75%, VA-ECMO: 87%; p = 0.22). The causes of death are delineated in Table 2. Among patients with STEMI presentation, ECPELLA was associated with lower mortality (HR 0.20 [0.06–0.66], p = 0.009), which was not different from patients without STEMI (HR 0.50 [0.22–1.11], p = 0.09, p for interaction = 0.49). Similarly, there was no interaction between SAVE score and mortality benefit with ECPELLA (SAVE less than −8: HR 0.47 [0.21–1.04], p = 0.063; SAVE greater than or equal to −8: HR 0.65 [0.27–1.54], p = 0.33, p for interaction = 0.45).

Table 2.: Causes of Death

Figure 1.: A: 30 day and (B) 1 year Kaplan–Meier survival estimates demonstrating statistically significant and maintained improved cumulative survival for the ECPELLA strategy as compared with VA-ECMO strategy. VA-ECMO, venoarterial extracorporeal membrane oxygenation.

Secondary Outcomes

There were no statistically significant difference between ECPELLA and VA-ECMO in terms of length of time on VA-ECMO (ECPELLA: 144 hours [94–186], VA-ECMO: 149 hours [73–219]; p = 0.57), the need for dialysis during hospitalization (22% vs. 27%; p = 0.69), incidence of stroke (n = 2 [5.6%] vs. n = 3 [10%]; p = 0.65) or transient ischemic attack (2.8% vs. 0%; p = 1.0), minor bleeding (11% vs. 10%; p = 1.0), major bleeding (36% vs. 33%; p = 1.0), or clinical hemolysis (22% vs. 27%; p = 0.78; Table 3).

Table 3.: Clinical Outcomes and Complications

Support and Weaning

Inotropic score was higher in VA-ECMO compared with ECPELLA on day 2 (ECPELLA: 0 [0–4], VA-ECMO: 11 [0–15]; p = 0.001) and day 3 (ECPELLA: 0 [0–2], VA-ECMO: 4 [0–16]; p = 0.02; Figure 2). Of the 36 patients that survived, the method of support 24 hours postdecannulation was reviewed: In the VA-ECMO cohort (n = 16), a majority (56%) required another form of mechanical cardiac support (n = 9; 89% IABP use; 11% temporary RVAD with oxygenator), 31% used inotropes only, and 13% required no support. In the ECPELLA cohort (n = 20), only 20% needed another form of mechanical cardiac support (n = 4; 50% Centrimag, 50% Heartmate II), while 55% maintained on Impella ± inotropes, and 10% inotropes alone. Three patients, all within the ECPELLA cohort, required recannulation for ECMO after their initial episode of VA-ECMO: Two individuals were converted to VV-ECMO for persistent lung failure after biventricular cardiac recovery, and one individual had severe bleeding at their axillary artery cannulation site, requiring new cannulation via the right femoral artery. Bridge to recovery, although not statistically different, was numerically almost double in the ECPELLA cohort as compared with the VA-EMCO cohort (40% vs. 22%; p = 0.18). Bridge to VAD was more prevalent in the ECPELLA group as compared with VA-ECMO (33% vs. 13%; p = 0.60). Specifically, with ECPELLA, two patients received Heartmate II devices and two patients received Centrimag (ultimately transitioned to Heartmate II devices), while with VA-ECMO, one patient received a temporary RVAD with oxygenator, but ultimately expired (Table 4).

Table 4.: Support and Weaning Outcomes

Figure 2.: Box-Whisker representation of differences in inotropic score for the first 3 days. A box represents the interquartile range (25th–75th percentile). Center line denotes the median score. There is a statistically significant decrease in the inotropic score after day 1 with the ECPELLA as compared with the VA-ECMO. VA-ECMO, venoarterial extracorporeal membrane oxygenation.

Hemodynamics

There were no significant statistical differences in the precannulation, postcannulation, and postdecannulation invasive hemodynamics (pulmonary artery systolic and diastolic pressures, pulmonary capillary wedge pressures) between the two cohorts. Notably, a numerically greater trend in the improvement of the CI in the ECPELLA cohort as compared with VA-ECMO alone was observed from precannulation (1.7 L/min × m² [1.1–2.5] vs. 2.3 L/min × m² [1.6–3.0]; p = 0.084) to 24 hour postdecannulation (3.3 L/min × m² [2.9–3.4] vs. 2.8 L/min × m² [1.9–3.3]; p = 0.244; Figure 3).

Figure 3.: Change in cardiac index in relation to VA-ECMO cannulation and decannulation. A trend in greater improvement of cardiac index is noted in the ECPELLA cohort as compared with the VA-ECMO cohort. VA-ECMO, venoarterial extracorporeal membrane oxygenation.

Discussion

We describe the largest US experience comparing VA-ECMO with and without simultaneous Impella support for the treatment of refractory CS.^28,29 In this cohort, we show that the addition of Impella to VA-ECMO is associated with improved survival of patients with refractory CS, decreasing need for inotropic support, without increasing complication rates.

Refractory CS can occur after cardiac surgery, as an exacerbation of chronic heart failure with reduced ejection fraction, or from de novo heart failure, most commonly caused by acute coronary syndromes. In our series, approximately 46% of cases were associated with acute coronary syndrome of which about 44% received revascularization. This is in keeping with previous reports which suggest that up to 50% of CS is because of some form of acute coronary syndrome.¹ Typically, in the setting of CS and ACS, IABPs were the preferred mode of mechanical support. However, in the contemporary IABP-SHOCK II trial, mortality was not improved by IABP and remained high at approximately 40% despite the use of a combination of inotropes, vasopressors, and IABP.³⁰ With the advent of Tandem heart (Cardiac Assist, Pittsburgh, PA) and Impella, percutaneous mechanical circulatory support devices can now be safely and quickly implanted through peripheral vessels without surgical intervention. Nevertheless, in the current era, CS mortality remains extremely high. In fact, even though both Impella and Tandem Heart are capable of achieving superior hemodynamics compared with IABP in CS, no survival benefit has yet been demonstrated in randomized controlled trial.^27,31,32 However, survival benefit with early initiation of support with Impella heart pumps in the setting of refractory CS complicating an AMI has been demonstrated in small single and multicenter observational studies,^33,34 as well as in large multicenter observational studies.^27,35 Furthermore, the use of early VA-ECMO has shown to reduce in-hospital mortality, and increase 30 day and 1 year survival in a recent small retrospective observational study.³⁶ However, this study enrolled a heterogeneous patient population with a lower risk profile on admission compared with our patients.

Despite being able to offer >5 L/min of oxygenated blood, VA-ECMO is limited in that it increases afterload, causing LV dilation, myocardial ischemia, elevated pulmonary pressures, blood stasis, and potential thrombus formation.^4–7 Further, the increase in afterload with retrograde blood flow into the aorta from the arterial inflow cannula opposes the central tenets of CS management which hinges on decreasing afterload and resting the myocardium. Prior surgical and animal studies have established that during cardiopulmonary bypass in the operating room, the use of left ventricular venting can decompress the chamber and helps prevent chamber dilation, stasis, and increased afterload.^37–42 Along these lines, a study combining the use of an IABP with VA-ECMO suggested that mechanical afterload reduction may be associated with improved in-hospital outcomes with low risk of complications¹⁶; however, a recent large pooled analysis of 1,517 patients demonstrated that there was no survival benefit with the routine combination of IABP and VA-ECMO.¹⁵ Pappalardo et al.²⁴ demonstrated that in a retrospective multicenter cohort of heterogenous European patients with CS treated with VA-ECMO plus Impella, this combination resulted in lower in-hospital mortality and higher rates of bridging to recovery or next therapy.

Contrary to the European experience, our cohorts had relatively similar baseline characteristics, and a unified decision for the initiation or escalation of mechanical support was made by a Shock Team rather than a single clinician.²⁴ The most significant finding in our study was that 30 day cumulative survival was greater in the ECPELLA versus VA-ECMO and extended to 1 year. Notably, the HR for overall mortality (ECPELLA versus VA-ECMO) was 0.52 (0.29–0.93; p = 0.027) suggesting a 50% mortality reduction from the addition of Impella to VA-ECMO. ST-elevation myocardial infarction and SAVE score did not significantly impact the benefit seen with ECPELLA, as evident by the lack of statistical interaction with respect to overall mortality. Although the mortality difference between the cohorts we report is unique, our overall cumulative survival rate is approximately 32% and 22% at 30 days and 1 year, respectively, for the entire cohort. These rates are similar to previously reported figures,^5,6 despite the fact that prior studies excluded patients with specific characteristics such as a postcardiotomy state or those who had previous CPR⁴²; conversely, the current study provides a real-world cohort of refractory CS to allow for an “all-comers” analysis. Further, the definition of “refractory” or “CS” is variable from study to study and thus the inclusion or exclusion of certain patients based on hemodynamics and clinical evaluations in these studies may alter survival rates; hence, we attempted to provide the most contemporary and meaningful definition of refractory CS to allow for some degree of applicability.

Nevertheless, we did observe a significant reduction in all-cause mortality at 30 days in the ECPELLA cohort. We believe these findings are related to several major observations. First and foremost, we report that only 58% of VA-ECMO patients (n = 21) had some form of venting (±IABP), as compared with 100% of the ECPELLA cohort (n = 30) which had Impella ± surgical vent (±IABP) supporting the association of left ventricular decompression and survival in our study. Along with the higher proportion of venting, the ECPELLA cohort demonstrated a numerical increase in CI from precannulation to postcannulation, and at decannulation as compared with the VA-ECMO cohort, despite using less inotropes. Similarly, a retrospective cohort of 121 patients reviewed for left ventricular distension while on VA-ECMO support revealed that the degree of clinically-defined left ventricular distention is inversely and significantly related to the degree of myocardial recovery, suggesting that lower degrees of distension through left ventricular decompression may result in improved cardiac function.²⁹ Further, in the present report, the use of the ECPELLA strategy resulted in more patients being successfully weaned from VA-ECMO (70% vs. 44%; p = 0.048). In contrast, over half of those who were decannulated in the VA-ECMO cohort required secondary mechanical support. Although not statistically different, over 40% of patients in the ECPELLA cohort recovered myocardial function as opposed to only 22% in the VA-ECMO cohort. Taken together, these data suggest that the addition of Impella to VA-ECMO may promote myocardial recovery and translate to improved survival. Second, ECPELLA patients were more likely to have percutaneous support placed before or during VA-ECMO initiation (n = 26, 87%), thus resulting in early unloading. Correspondingly, a study of 287 patients with CS treated with Impella and stratified by time to support demonstrated that a shorter time to unloading correlated with improved survival.³⁵ We hypothesize that a similar benefit is observed in patients who have early and effective unloading with combination Impella and VA-ECMO. Third, as noted above, the maximal inotropic score after day 1 was significantly higher in the VA-ECMO compared with the ECPELLA group. This suggests that the use of Impella may provide improved hemodynamic support as compared with the use of inotropes, while simultaneously reducing the use and harmful effects of these drugs (arrhythmias, progressive renal dysfunction, increase myocardial oxygen demand and ischemia, increased risk of cardiac death).^43–45 Finally, the use of the ECPELLA strategy has the added clinical utility of providing a transitional strategy that allows earlier cessation of VA-ECMO. Indeed, a recent large meta-analysis has shown that VA-ECMO is associated with substantial morbidity.⁴⁶ Furthermore, an analysis of outcomes of 4,658 patients from the national inpatient database in Japan supported with VA-ECMO for CS found excessive in-hospital mortality of 73.6% (77.1% in patients who sustained cardiac arrest [CA]). Of those, 40.3% expired after decannulation suggesting subsequent complications or lack of total native heart recovery.⁴⁷

In terms of safety, there were no differences in major complications between ECPELLA and VA-ECMO alone. Any degree of bleeding, clinical hemolysis, stroke, and the need for dialysis were not different in either cohort. This confirms previous observations of comparable safety profiles in a small series comparing surgical decompression to Impella.²⁸ We postulate that the low rate of observed complications is related to the adoption of contemporary management strategies for CS.^48–50 In essence, our Shock Team has become seasoned in its approach and so to have the individual team members, including cardiothoracic surgery and interventional cardiology, who have become adept in large vessel access and closure, Impella placement and repositioning, and monitoring for and managing bleeding and hemolysis. Similarly, European experience has also found a favorable safety profile using a combination of Impella and VA-ECMO employing similar CS management strategies.²⁴

Limitations

Although this is the largest US-based study of the ECPELLA strategy versus VA-ECMO for the treatment of CS, we acknowledge that the sample size was relatively small, thus limiting our ability to propensity match patients; however, baseline characteristics were relatively similar between the two groups. In addition, our patients were retrospectively compared, thus introducing associated biases. Although we attempted to clearly define refractory CS, the final cohort was likely heterogeneous in terms of overall clinical picture. Finally, our surgical and interventional cardiology experience is extremely mature with the use of a Shock Team approach to treat refractory CS, thus enabling the use of advanced hemodynamic support devices and reperfusion strategies, which makes generalizability of our results limited.

Conclusions

This study suggests that the addition of Impella to VA-ECMO is associated to improved survival of patients with refractory CS. Randomized controlled trials are required to confirm these findings.