Venoarterial Extracorporeal Membrane Oxygenation With Concomitant Impella Versus Venoarterial Extracorporeal Membrane Oxygenation for Cardiogenic Shock

Abstract

There are contrasting data on concomitant Impella device in cardiogenic shock patients treated with venoarterial extracorporeal membrane oxygenation (VA ECMO) (ECPELLA). This study sought to compare early mortality in patients with cardiogenic shock treated with ECPELLA in comparison to VA ECMO alone. We reviewed the published literature from 2000 to 2018 for randomized, cohort, case-control, and case series studies evaluating adult patients requiring VA ECMO for cardiogenic shock. Five retrospective observational studies, representing 425 patients, were included. Venoarterial extracorporeal membrane oxygenation with concomitant Impella strategy was used in 27% of the patients. Median age across studies varied between 51 and 63 years with 59–88% patients being male. Use of ECPELLA was associated with higher weaning from VA ECMO and bridging to permanent ventricular assist device or cardiac transplant in three and four studies, respectively. The studies showed moderate heterogeneity with possible publication bias. The two studies that accounted for differences in baseline characteristics between treatment groups reported lower 30 day mortality with ECPELLA versus VA ECMO. The remaining three studies did not adjust for potential confounding and were at high risk for selection bias. In conclusion, ECPELLA is being increasingly used as a strategy in patients with cardiogenic shock. Additional large, high-quality studies are needed to evaluate clinical outcomes with ECPELLA.

Despite advances in the delivery of cardiovascular care, cardiogenic shock (CS) is associated with 30–45% in-hospital mortality in the contemporary era.^1,2 Recent advances in the field of acute mechanical circulatory support (MCS) have resulted in a paradigm shift in the management of these patients.³ Venoarterial extracorporeal membrane oxygenation (VA ECMO) is often used for CS due to etiologies such as acute myocardial infarction (AMI), postcardiotomy shock, acute myocarditis, and end-stage heart failure.^1,4–6 Venoarterial extracorporeal membrane oxygenation has multiple theoretical advantages that include rapid bedside access with/without fluoroscopy, high cardiac output support of 3–5 L, ability to support both cardiac and pulmonary function and relatively lower patient costs.⁷ However, central or peripheral VA ECMO cannulas do not adequately unload the left ventricle (LV) resulting in increased afterload, “mixing” syndrome and hydrostatic pulmonary edema.⁸ Furthermore, use of peripheral (or even central) VA ECMO is associated with stagnation of blood in the hypokinetic LV, which may result in formation of an LV thrombus.⁹ Several methods to unload the LV during VA ECMO have been described including surgical venting, atrial septostomy, or concomitant MCS devices such as the intra-aortic balloon pump (IABP) or percutaneous LV assist devices (LVADs).

Prior work focusing on the use of dual MCS devices has predominantly involved the IABP.^8,10 The Impella percutaneous LVAD (Abiomed, Danvers, MA) was recently approved for use in the United States and has been adopted for use in patients with CS.³ The Impella device can provide varying cardiac output support of 2.5–5.0 L by directly unloading the LV across the aortic valve. The results of the use of VA ECMO concomitantly with Impella (ECPELLA) have been limited to case reports and small single-center studies. In light of the multiple recent studies demonstrating contrasting results,^11–15 we sought to review the available evidence of the impact of ECPELLA versus VA ECMO on early mortality in patients with CS. Our primary hypothesis was that use of ECPELLA is associated with decreased early mortality compared with VA ECMO.

Materials and Methods

Data Sources and Search Strategies

A comprehensive search of several databases from January 1, 2000, to May 13, 2018, English language, was conducted. The databases included Ovid MEDLINE Epub Ahead of Print, Ovid Medline In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus. The search strategy was designed and conducted by a medical librarian with input from the study’s first author. Controlled vocabulary supplemented with keywords was used to search for mortality outcomes in CS needing VA ECMO with the concomitant use of Impella in adults. The detailed search strategy is presented in Supplementary Table 1, https://links.lww.com/ASAIO/A447. The resultant abstracts were screened by two independent reviewers (P.A., S.A.). All references of included studies were evaluated for additional studies. Study inclusion was based on the consensus of the two reviewers. A third independent reviewer (S.V.) served as the referee in case of disagreement between the first two reviewers in conjunction with the first author (S.V.). The search strategy and reporting were performed using strengthening the reporting of observational studies in epidemiology guidelines.

Inclusion and Exclusion Criteria

Studies that reported early mortality in adult patients (> 18 years) with CS necessitating VA ECMO were included. The primary outcome was early mortality defined as mortality during intensive care unit stay, hospital stay or mortality less than or equal to 30 days. Literature from human studies and of case-control, cohort, and randomized trial study designs were included. In studies reporting outcomes in unselected shock patients on VA ECMO, only studies for which a 2 × 2 table could be constructed/abstracted between Impella use and mortality were included. Abstracts presented at professional societal meetings were excluded. Studies designed as case reports/series, systematic or narrative reviews, pediatric or animal studies, and studies without relevant outcomes were excluded. If multiple studies were published by the same group of authors over the same study duration, only a single study with relevant outcomes was included. Data abstracted included study year, population, location, type of study, MCS-related parameters, and clinical outcomes. Quality was assessed using the Newcastle Ottawa Scale (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp).

Statistical Analysis

Heterogeneity among the studies was estimated using the I² statistic as described by Higgins et al.¹⁶ Publication bias was estimated by visual inspection of funnel plots for asymmetry. Given the high risk of bias, evidence of publication bias and statistical heterogeneity between the published studies, a meta-analysis was not performed. Therefore, available evidence was summarized qualitatively. Odds ratios (ORs) and 95% confidence intervals (CIs) for early mortality between the two treatment groups for each study were calculated using reported numbers of events.

Results

The search strategy identified 392 abstracts, of which five studies, representing 425 patients met the inclusion criteria (Figure 1 and Table 1).^11–15 All studies were retrospective cohort studies evaluating early mortality in patients with refractory CS. Heterogeneity between studies was moderate between studies (I² = 66%), and funnel plot suggested possible publication bias (Figure 2). Quality of the studies, as assessed using the Newcastle Ottawa Scale showed a low risk of bias (Supplementary Table 2, https://links.lww.com/ASAIO/A447). All studies were retrospective observational studies; there were no randomized trials on this subject. Three studies were from the United States and two from Europe. Cardiogenic shock definition was specified in only in three studies (60%) and was defined using a combination of hemodynamic targets, end-organ injury, and increasing or maximal doses of inotropes and vasopressors (Table 1).^11,12,14 Akanni et al.¹¹ evaluated 29 patients with unselected CS who were converted to ECPELLA from either VA ECMO (n = 14) or Impella (n = 15) monotherapy and compared them to patients receiving VA ECMO during the same time period. The percutaneously implantable Impella LP 2.5 or CP systems were used in all patients. The VA ECMO circuit used comprised a Quadrox iD oxygenator (Maquet, Wayne, NJ), the Rotaflow centrifugal pump (Maquet, Wayne, NJ), and SMART-coating tubing (Sorin, Mirandola, Province of Modena, Italy) with peripheral cannulation for arterial cannula (15–17 Fr). Patients who received Impella monotherapy were predominantly in CS due to AMI (73%) whereas the patients who received initial VA ECMO monotherapy had postcardiotomy CS (15%).¹¹ In the study by Mourad et al.¹², 42 AMI patients with CS, were treated with VA ECMO (n = 23) and Impella (n = 19) monotherapy initially with a subsequent conversion to ECPELLA strategy. Impella 5.0, CP, and 2.5 were used in 7, 7, and 5 patients, respectively. There were no differences in percutaneous coronary intervention (PCI) procedural characteristics or outcomes between the groups.¹² Pappalardo et al.¹³ evaluated 34 CS patients treated with ECPELLA in two cardiac intensive care units in Italy and Germany and compared them to propensity-matched controls on VA ECMO. The percutaneously implantable Impella LP 2.5 or CP systems were used in all patients. A centrifugal pump with coated polymethylpentene oxygenator with output cannulas 21–29 Fr and inlet cannulas of 15–19 Fr were used for the VA ECMO circuit. Patients in the control group of this study had higher rates of cardiac arrest and preimplantation cardiopulmonary resuscitation as compared with the ECPELLA cohort.¹³ In a cohort of 66 patients, Patel et al.¹⁴ noted ST-elevation AMI and PCI to be more common the ECPELLA (n = 30) as compared with the VA ECMO only (n = 36) cohort, but there were no differences in rates of cardiopulmonary resuscitation. Impella 5.0, CP, and 2.5 were used in 4, 24, and 2 patients, respectively. 15–19 Fr arterial cannulas and 23–25 Fr multihole venous attached to an external centrifugal pump and an oxygenator were used for the VA ECMO circuit. Around 20–30% in both cohorts also received additional venting with IABP or surgical vent. Tepper et al.¹⁵ compared 23 ECPELLA patients (initial Impella, 14; initial VA ECMO, 9) to 22 VA ECMO patients with surgical venting. Impella 5.0, CP, and 2.5 were used in 9, 7, and 7 patients, respectively. Patients who received ECPELLA were predominantly in CS due to AMI (39%) whereas the patients who received VA ECMO predominantly had postcardiotomy CS (50%).

Table 1.: Characteristics of Included Studies

Figure 1.: Literature search strategy.

Figure 2.: Duval and Tweedie Trim and Fill Method funnel plot analysis of publication bias. OR, odds ratio; SE, standard error.

In total, 312 patients (73.4%) were treated with only VA ECMO and 113 patients (26.6%) with ECPELLA strategy. Median age across studies varied between 51 and 63 years with 59–88% patients being male (Table 2). A majority of the VA ECMO devices were peripherally inserted among the studies that reported this information (Table 3). Concomitant PCI was used in all patients in the study evaluating acute myocardial infarction,¹² and a majority of the patients in other cohorts as noted in Table 3. Use of ECPELLA strategy was associated with improvements in systolic and diastolic pulmonary arterial pressures (Akanni et al.¹¹), improvements in central venous pressures (Tepper et al.¹⁵), decrease in transaminitis (Tepper et al.¹⁵), decreased total vasoactive medication requirements (Akanni et al.¹¹), higher VA ECMO weaning (Pappalardo et al.¹³, Patel et al.¹⁴, and Tepper et al.¹⁵) and greater bridging to recovery (Patel et al.¹⁴) or bridge to permanent therapies such as durable LVAD and cardiac transplant (Akanni et al.¹¹, Pappalardo et al.¹³, Patel et al.¹⁴, and Tepper et al.¹⁵) (Table 2). Renal failure requiring hemodialysis, major bleeding requiring transfusion, limb ischemia, and hemolysis were reported with varying incidence across the studies (Table 3).

Table 2.: Baseline Characteristics and Outcomes of ECPELLA Versus VA ECMO in Cardiogenic Shock

Table 3.: Procedural Characteristics and Complications of ECPELLA Versus VA ECMO in Cardiogenic Shock

Mortality

The overall mortality was high in the cohorts with ECPELLA therapy (54%) versus VA ECMO alone (55.1%). Mortality among the individual studies varied between 32 and 78% for ECPELLA and 43–80% for VA ECMO patients (Table 2). Only two studies adjusted for baseline confounding using multivariable regression or propensity-matching analysis.^13,14 Both these studies reported a lower mortality in the ECPELLA cohort as compared with the VA ECMO cohort—48% vs. 74%, p = 0.04 in propensity-matched patients (Pappalardo et al.¹³) and adjusted hazard ratio 0.39 (95% CI: 0.19–0.81, p = 0.01) (Patel et al.¹⁴). In the study by Mourad et al.¹², early mortality was higher in the ECPELLA cohort (8/10 deaths [ECPELLA] vs. 6/16 deaths [VA ECMO]) (OR, 6.67; 95% CI: 1.05–42.4). The remaining two studies reported no difference in early mortality between the two treatment groups (Figure 3).

Figure 3.: Early Mortality in patients with ECPELLA versus venoarterial ECMO for cardiogenic shock. *Adjusted hazard ratio (95% CI) presented for Patel et al. ¹⁴. CI, confidence interval; ECMO, extracorporeal membrane oxygenation, ECPELLA, venoarterial extracorporeal membrane oxygenation with concomitant Impella.

Discussion

In this first systematic review evaluating ECPELLA versus VA ECMO for CS, the use ECPELLA was noted in 27% of the patients. Studies encompassed a variety of CS etiologies including AMI, postcardiotomy shock, and mixed etiologies. Overall mortality in the total study was 54.8% consistent with existing literature, with studies that adjusted for baseline confounding noting a lower mortality in the ECPELLA cohort. Both strategies were associated with a high rate of complications including major bleeding, transfusions, limb ischemia, and acute kidney injury.

Dual Mechanical Circulatory Support in Cardiogenic Shock

In patients needing VA ECMO for CS, LV distension is seen in nearly 30–40% of the patients that is associated with pulmonary edema, ventricular arrhythmias and stagnation of blood in the LV.¹⁷ The use of a second MCS device to unload the LV during the use of VA ECMO offers a theoretical advantage of LV unloading and thereby preventing distension, decreasing myocardial oxygen demand and prevention of hydrostatic pulmonary edema.⁸ Traditionally, the IABP has been used in these patients, but multiple studies did not show convincing evidence of improved outcomes of concomitant IABP use.^10,18 More recent data has shown that concomitant IABP decreases hydrostatic pulmonary edema thereby potentially aiding earlier extubation and rehabilitation.⁸ However, the magnitude of hemodynamic support generated by an IABP is directly related to LV cardiac output, which further decreases its efficacy in CS.¹⁹ The axillary rotary pump of the Impella on the other hand, is capable of generating 2.5–3.5 L of flow associated with improvements in coronary perfusion.¹⁹ As compared with the IABP, the Impella has been associated with improved hemodynamic end-points without any differences in mortality in patients with CS complicating acute myocardial infarction.²⁰ In patients with CS, the ECPELLA was associated with significant decreases in central venous pressures, pulmonary pressures and vasoactive medication requirements as compared to on VA ECMO.^11,15 The use of a ECPELLA was associated with greater ECMO weaning and bridging to permanent therapies (durable LVAD or cardiac transplantation) in the studies that reported this information.^11,13–15

Outcomes in ECPELLA Versus VA ECMO

Due to high baseline confounding and inadequate adjustment for severity of illness, this study was unable to directly compare the early mortality in the two cohorts. Certain essential factors relating to study design, patient selection and baseline variables merit further discussion. First, the inclusion of heterogeneous cohorts of CS etiology may confound comparison since AMI and postcardiotomy shock likely represent systematically different patients.¹⁸ In contrast to CS due to AMI or heart failure, patients with postcardiotomy shock often have complex hemodynamics due to a combination of cardiogenic and vasodilatory phenotypes.²¹ Furthermore, in the studies that reported this information, bias by indication was noted with higher use of initial Impella and VA ECMO monotherapy in AMI and postcardiotomy shock respectively. Second, there was incomplete reporting on the inclusion of cardiac arrest patients who received resuscitation involving VA ECMO. These patients have a significantly worse prognosis based on the duration and extent of cardiac arrest and cardiopulmonary resuscitation and form a distinct cohort from CS without cardiac arrest.²² As noted in the study by Akanni et al.¹¹, patients with cardiac arrest were more common in the Impella to ECPELLA subgroup that had longer lengths of ICU stay suggestive of greater severity of illness. Third, it is likely that the hemodynamic benefits of the Impella could likely be off-set greater organ failure and microcirculatory dysfunction in CS in the cohort that predisposed these patients to receive dual MCS (confounding by indication).^1,13 Furthermore, the timing of Impella and VA ECMO support are important, as earlier unloading of the LV before development of vasoplegia and multiorgan failure has important prognostic implications.¹⁹ Lastly, the role of LV venting in patients without clinical or radiographic features of LV distension is less clear and the Impella does not appear to confer any clinically significant benefit to these patients.^15,17 In the studies that did adjust for baseline confounding, the use of ECPELLA strategy did demonstrate significantly lower mortality.^13,14 Further dedicated studies with adequate statistical power and adjustment are required to understand the role of ECPELLA in CS mortality. Future research on ECPELLA strategy should account for baseline differences by adjusting for confounding, preferably in a prospective fashion.¹⁴ Importantly, the clinical evolution of CS is significantly different in patients with AMI-CS and post-cardiotomy shock and these patient populations should ideally be assessed independently.¹⁸ Importantly, the indication for the placement of a secondary device (usually an Impella for patients on VA ECMO) should be defined by objective criteria, such as elevated wedge pressures, persistent pulmonary edema, inability to wean mechanical ventilation or objective echocardiographic criteria of LV distention.⁸ Ideally, hemodynamic variables using a pulmonary artery catheter should be used to document clinical improvement in addition to mortality outcomes.²³ Lastly, the ECPELLA strategy should be compared with the IABP or surgical/percutaneous venting of the LV to establish clear superiority before widespread clinical application.^18,24

Despite the stated potential benefits, the role of ECPELLA strategy for CS needs to be balanced against the risks of 1) blood transfusions for hemorrhage or hemolysis;²⁰ 2) vascular complications, such as retroperitoneal hemorrhage, distal limb ischemia, and arterial laceration from the use of large bore femoral cannulas; and 3) lack of conclusive information on mortality benefit. Furthermore, in the limited studies that reported this information, the ECPELLA strategy was associated with higher need for transfusion and acute kidney injury. It is possible that greater hemolysis and potential vascular complications from dual circulatory support may contribute towards the need for higher transfusions.⁶ Despite improved hemodynamics with the ECPELLA strategy the effect of vascular complications and prerenal hypoperfusion from bleeding may have resulted in higher acute kidney injury.⁶ Alternately, it is possible that in postcardiotomy patients, development of distributive shock may have resulted in a cardiorenal syndrome analogous to other vasoplegic shock states such as sepsis.^21,25,26 Careful patient selection remains key to the use of acute MCS in CS. In this review, studies used inconsistent definitions of CS and only a few studies defined “refractory” using objective hemodynamic data. Furthermore, the use of vasopressors and inotropes in CS remains empiric and unquantified in these studies, which could potentially influence the clinical outcomes in these populations.¹ Development of validated clinical cutoffs for vasoactive medication support to define refractoriness, early initiation of MCS to prevent acute organ failure and hemometabolic shock and prevention of vascular and bleeding complications remain clinical priorities for the management of refractory CS needing MCS.^1,3,19

Limitations

This study has important limitations. Detailed baseline data, specifically an evaluation of early revascularization, timing of MCS, and end-organ damage were not uniformly reported across studies. This study is limited in assessing the Impella and cannot comment on other percutaneous LVADs such as the TandemHeart (Cardiac Assist Inc, Pittsburgh, PA). The comparator group (VA ECMO only) had inconsistent use of alternate LV unloading devices such as the IABP or a surgical vent. Further data are needed to develop head-to-head comparisons of the various LV unloading strategies in patients receiving VA ECMO. Finally, this study evaluated early mortality only with limited insight into long-term survival and functional recovery in patients with CS.

Conclusions

In this review of 425 patients with CS requiring VA ECMO support, we noted an increasing use of ECPELLA strategy for CS. Noting the inherent limitations of this review, high-quality data using either randomization or propensity matching are needed to evaluate the incremental benefit of the ECPELLA strategy in the management of CS.

ACKNOWLEDGMENT

The authors thank Larry J. Prokop, MLS from the Mayo Clinic Libraries for his assistance with the literature search.