Concomitant implantation of Impella® on top of veno-arterial extracorporeal membrane oxygenation may improve survival of patients with cardiogenic shock
Abstract
Aims
Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) support stabilizes patients with cardiogenic shock. Despite improved oxygenation and peripheral circulation, LV unloading may be impeded due to the increased afterload, resulting in a failing static left ventricle and in high mortality.
Methods and results
We describe for the first time a large series of patients treated with the combination of VA-ECMO and Impella® compared with patients with VA-ECMO only. We retrospectively collected data on patients from two tertiary critical care referral centres. We enrolled 157 patients treated with VA-ECMO from January 2013 to April 2015: 123 received VA-ECMO support and 34 had concomitant treatment with VA-ECMO and Impella. A propensity-matching analysis was performed in a 2:1 ratio, resulting in 42 patients undergoing VA-ECMO alone (control group) compared with 21 patients treated with VA-ECMO and Impella. Patients in the VA-ECMO and Impella group had a significantly lower hospital mortality (47% vs. 80%, P < 0.001) and a higher rate of successful bridging to either recovery or further therapy (68% vs. 28%, P < 0.001) compared with VA-ECMO patients. A higher need for continuous veno-venous haemofiltration (48% vs. 19%, P = 0.02) and increased haemolysis (76% vs. 33%, P = 0.004) were reported in the study group due to higher survival. There was no difference in major bleeding rates between the two groups (VA-ECMO and Impella 38% vs. VA-ECMO 29%, P = 0.6).
Conclusions
Concomitant treatment with VA-ECMO and Impella may improve outcome in patients with cardiogenic shock compared with VA-ECMO only. Nevertheless, randomized studies are needed to validate these promising results further.
Introduction
Cardiogenic shock (CS) is associated with high mortality,1 although medical and non-medical treatment have improved over the last decades.2 Mechanical cardiac assist devices such as veno-arterial extracorporeal membrane oxygenation (VA-ECMO) have increasingly been used in critically ill patients suffering from CS.3 VA-ECMO has shown a survival benefit in cardiopulmonary resuscitation (CPR)4 and is used as a bridge to recovery or a bridge to decision, with improved survival in case series.5 Nevertheless, no trials exist providing controlled evidence of this approach.
Among patients with CS, acute implantation of mechanical circulatory assist devices is associated with poor outcome and a high complication rate.6 With respect to VA-ECMO, there is a haemodynamic pitfall that can potentially impair outcome in patients treated with VA-ECMO: intrinsic to this type of circulatory support is its limited capability to unload the left ventricle. Indeed, it even increases systemic afterload secondary to retrograde blood flow.7 This dilemma predominantly relates to patients suffering from a dysfunction that directly affects the myocardium, as in acute cardiac pump failure due to myocardial infarction (MI) or myocarditis.7 A lack of LV unloading combined with an elevated afterload can cause ventricular distension with a high risk of consecutive myocardial ischaemia,8 pulmonary oedema,9, 10 and, finally, LV and pulmonary thrombosis due to flow stasis.11 These complications increase mortality.12
Current strategies of LV unloading while on VA-ECMO include central percutaneous cannulation of the left atrium (LA) and the axillary artery13 or the left ventricle,7 as well as utilization of an intra-aortic balloon pump (IABP).14 A case series showed improved haemodynamics as well as LV unloading after adding an IABP to a VA-ECMO system, but no direct controls were presented.15 Impella® (ABIOMED), which are percutaneously placed inside the left ventricle, are used as cardiac assist devices in CS as well as high-risk coronary interventions.6, 17 Their potential to unload the left ventricle effectively in patients treated with VA-ECMO was described in single case studies18, 19 and in one case series involving five VA-ECMO and Impella cases.20 In addition to providing unloading of the left ventricle, it has the haemodynamic effect of increasing the net forward flow.
We describe here the largest multicentre, propensity-matched study of patients presenting with CS treated with the combination of Impella and VA-ECMO as compared with patients treated with VA-ECMO alone in view of outcome.
Methods
The present study is in compliance with the Declaration of Helsinki. Collection and analysis of data were carried out in agreement with Italian and German data protection laws.
Design of the study
We report patients with CS in need of specialized treatment. We compared patients receiving a combination of VA-ECMO and Impella implanted simultaneously with patients treated with VA-ECMO. To adjust for concomitant diseases, risk factors, and shock severity, we performed additional propensity score matching.
Patient assessment
Between January 2013 and April 2015, 34 patients with severe refractory CS were treated with a combination of mechanical cardiac assist devices consisting of an Impella 2.5 or Impella CP (ABIOMED, Danvers, USA) LV pump and VA-ECMO in the Cardiac Intensive Care Unit of San Raffaele Scientific Institute (Milan, Italy) and the Cardiac Intensive Care Unit of the University Heart Centre Hamburg Eppendorf (Hamburg, Germany). All consecutive patients with implantation of Impella and VA-ECMO were included in this study. Crucial for the inclusion in the study was the indication for VA-ECMO implantation for LV forward failure. The decision for an additional implantation of Impella was undertaken as the attending physician recognized signs of echocardiographic, radiological, and clinical signs of impaired LV unloading or LV stasis (stone heart, pulmonary oedema, impending clotting on the left ventricle, significant aortic regurgitation). Baseline characteristics including type of CS as well as haemodynamic, metabolic, and laboratory data together with documented complications were compared with a contemporary control group of 123 CS patients treated with VA-ECMO only, which did not receive LV unloading via a percutaneous LV assist device (LVAD). These patients were matched for gender, age, baseline lactates and pH, pre-implant CPR, PCI, ST-segement elevation myocardial infarction (STEMI), and concomitant IABP support. Data acquisition was done retrospectively by means of electronic patient files and documentation system search, and included patients from January 2013 to April 2015.
Implantation of veno-arterial extracorporeal membrane oxygenation device
The VA-ECMO circuit set-up included a centrifugal pump and a coated polymethylpentene oxygenator. All patients were intubated and received mechanical ventilation at the time of VA-ECMO implantation. As for peripheral cannulation, outlet cannulas ranged from 21 to 29 French, and inlet cannulas from 15 to 19 French. A distal perfusion cannula (range 5–8 French) was placed whenever possible in the superficial femoral artery to prevent leg ischaemia. VA-ECMO flow was set at the maximal speed in order to produce reduction of the metabolic picture of malperfusion.
Implantation of Impella
Impella 2.5 or Impella CP were implanted20 through the femoral artery and placed via a retrograde approach through the aortic valve into the left ventricle. Impella 2.5 or Impella CP were left running at P8 speed in order to produce a forward flow of 2.0 L without complications. Pump position was routinely checked with echocardiography and chest X-ray, display data being not fully reliable in this setting.
In the case of suction or decreased blood flow, echocardiographically guided repositioning of the Impella was performed and pump speed was optimized.
Explantation of devices
As the signs of shock subsided and echocardiography showed myocardial recovery, VA-ECMO flows were progressively reduced and weaning was performed as soon as possible. It was targeted that the Impella was left implanted until the patient had further improvement in his clinical status. This was performed whenever possible without clinical necessity to remove the device earlier due to complications. All decisions to explant any device was done at the discretion of the attending physician.
In patients who were showing no signs of myocardial recovery, but who were indeed neurologically intact, a team discussion drove further therapy to LVAD or heart transplantation.
Concomitant medications
Patients received anticoagulation during mechanical support with either unfractionated heparin or bivalirudin titrated to an activated partial thromboplastin time (aPTT) between 45 and 60 s.
The purge fluid of the Impella pump also contained unfractionated heparin (either 25 or 50 U/mL) according to clinical need. Antiplatelets were administered according to clinical need.
General medical treatment with inotropes and other supportive therapies was performed according to clinical need and with the guidance of current guidelines or standard operation protocols of both Cardiac Intensive Care units.
Outcomes
The primary outcome of the study was hospital mortality; secondary outcomes included the rate of bridging to next therapy (LVAD, heart transplantation) or myocardial recovery, and complications.
Definitions of bleeding, haemolysis, and access site complications
Major bleeding included all cases of intracranial, retroperitoneal, intraocular, and retropharyngeal bleeding, bleeding of the cannulation site requiring either radiological or surgical intervention, cannulation site haematoma >5 cm, a decrease in haemoglobin serum value >4 g/dL in the absence of identification of the bleeding site, or >3 g/dL in the case of an identified bleeding site, and bleeding with the need for transfusion of at least three red blood cells units.
Minor bleeding included all cases of bleeding not meeting the criteria for the definition of major bleeding.
Haemolysis was defined as an increase in lactate dehydrogenase (LDH) serum levels above 1000 U/L associated with an increase in plasma free haemoglobin above the upper limit of the laboratory range in at least two consecutive blood samples within 24 h.
The SAVE (Survival After Veno-arterial ECMO) score was calculated as indicated by Schmidt and colleagues.21
Access site complications
Access site complications are defined as all ischaemic events of the lower limb distal to the cannulation site as well as infections of the device insertion site ultimately caused by the implantation of the device.
Statistical methods
Categorical variables are reported as the absolute number and percentage, whereas continuous variables are expressed as the median with interquartile range (25th–75th percentile) according to their distribution. Normality was examined by means of both Shapiro–Wilk and Kolmogorov–Smirnov tests. The prevalence of dichotomous covariates was compared between cases (VA-ECMO and Impella) and controls (VA-ECMO) using a χ2 test (or a Yates' χ2 test when at least one cell of the table has an expected count smaller than five), while a standard two-sample t-test (or a Wilcoxon test for non-normal distribution) was used to compare continuous covariates.
To estimate the probability of treatment assignment conditional on observed baseline covariates, namely the propensity score, we used a logistic regression to regress receipt of an Impella device on the following baseline characteristics, which presented a clinical and/or statistical relevance: age, gender, baseline lactates and pH, pre-implant CPR, PCI, STEMI as the cause of the arrest, and concomitant IABP support. We matched cases and control patients (2:1 matching) who share a similar value of the propensity score using the greedy algorithm using the ‘nearest neighbour matching’ criteria.22
The absolute standardized differences were used to evaluate the propensity score model performance. A value of the absolute standardized difference <0.2 identifies a smaller effect of the covariate on the treatment assignment.23
Multiple logistic regression models were carried out to identify the possible predictors of hospital mortality in the propensity score-matched sample.
Statistical significance was set at the two-tailed 0.05 level for testing the hypothesis. Unadjusted P-values are reported throughout. Data were analysed using Microsoft Excel 2007 (Microsoft Office 2007, Redmond, WA, USA) and SAS 9.2 (SAS Institute Inc., Cary, NC, USA).
Results
Baseline demographics of all patients included
A total of 157 patients treated for severe CS at both centres were included in this study; of these, 123 (78%) received VA-ECMO support and 34 had concomitant treatment with VA-ECMO and Impella (22%). Median age was 55 (46–64) years, and 83% of patients were male. Baseline characteristics of the overall population are shown in Table 1. At baseline, patients in the control group presented a higher rate of pre-implant CPR as compared with patients in the study group (70% vs. 41%, P = 0.002), and lower pH [7.16 (6.95–7.37) vs. 7.36 (7.08–7.41), P = 0.02].
| Parameter | Total (n = 157) | ECMO + Impella (n = 34) | ECMO (n = 123) | P-value | Absolute standardized differencea |
|---|---|---|---|---|---|
| Age, years | 55 (46–64) | 54 (47–66) | 55 (45–64) | 0.9 | 0.0217 |
| Males, n (%) | 130 (83) | 28 (82) | 102 (83) | 0.9 | 0.0263 |
| CPR, n (%) | 100 (64) | 14 (41) | 86 (70) | 0.002 | 0.6101 |
| STEMI, n (%) | 85 (54) | 15 (44) | 70 (57) | 0.2 | 0.2622 |
| PCI, n (%) | 56 (36) | 16 (47) | 40 (33) | 0.10 | 0,2887 |
| pH | 7.23 (6.98–7.39) | 7.36 (7.08–7.41) | 7.16 (6.95–7.37) | 0.02 | 0.5087 |
| Lactates, mmol/L | 9.55 (4.40–15.35) | 8.96 (3.10–16.25) | 10.26 (4.97–15.24) | 0.4 | 0.1173 |
| Concomitant IABP, n (%) | 54 (34) | 8 (24) | 46 (37) | 0.10 | 0.2852 |
- CPR, cardiopulmonary resuscitation; IABP, intra-aortic balloon pump; STEMI, ST-segment elevation myocardial infarction.
- a 0–0.2 small effect size; 0.2–0.5 medium effect size; 0.5–0.8 large effect size; 0.8–1 very large effect size.
Overall outcomes
The overall hospital mortality was 73%. Notably, patients in the VA-ECMO and Impella group experienced a statistically significant lower hospital mortality (47% vs. 80%, P < 0.001), and a higher rate of successful bridging to either recovery or next therapy (68% vs. 28%, P < 0.001).
The bleeding rate was not different among the two groups; haemolysis occurred more frequently in patients treated with VA-ECMO and Impella (76% vs. 20%, P < 0.001), and the need for continuous veno-venous haemofiltration (CVVH) was similarly higher in the study population (53% vs. 23%, P = 0.001).
Both the duration of VA-ECMO treatment and the duration of mechanical ventilation were longer in patients treated with VA-ECMO and Impella [167 (72–286) vs. 61 (16–168), P < 0.001; and 163 (80–502) vs. 61 (13–192), P < 0.001, respectively].
Baseline demographics of propensity-matched cohort
The 2:1 propensity score matching identified 21 patients treated with VA-ECMO and Impella and 42 patients with VA-ECMO alone. The performance of the propensity score model is shown in Table 2, depicting baseline characteristics of the two populations, which entered the propensity matching. The means and the proportions of baseline covariates are similar between cases and controls (absolute standardized difference <0.2) in the propensity score-matched sample.
| Parameter | Total (n = 63) | ECMO + Impella (n = 21) | ECMO (n = 42) | P-value |
|---|---|---|---|---|
| Age, years | 53 (46–65) | 51 (47–61) | 54.5 (46–65) | 0.6 |
| Males, n (%) | 55 (87) | 18 (86) | 37 (88) | 0.5 |
| CPR, n (%) | 40 (63) | 12 (57) | 28 (67) | 0.5 |
| STEMI, n (%) | 30 (48) | 10 (48) | 20 (48) | 1 |
| PCI, n (%) | 27 (43) | 9 (43) | 18 (43) | 1 |
| pH | 7.27 (7.00–7.41) | 7.31 (7.08–7.39) | 7.27 (6.98–7.43) | 0.7 |
| Lactates, mmol/L | 9.02 (4.05–14.17) | 9.02 (4.60–11.00) | 9.03 (4.05–14.17) | 1 |
| Concomitant IABP, n (%) | 13 (21) | 6 (29) | 7 (17) | 0.3 |
- CPR, cardiopulmonary resuscitation; IABP, intra-aortic balloon pump; STEMI, ST-segment elevation myocardial infarction.
Outcomes of propensity-matched cohort
The reduction of hospital mortality and the higher rate of bridging to recovery or next therapy observed in the Impella group on the overall population was further confirmed in this propensity score-matched sample: hospital mortality was 48% in the study group vs. 74% in the control group (P = 0.04), and the rate of bridging to recovery or next therapy was 62% vs. 36%, respectively (P = 0.048) (Table 3).
| Parameter | Total (n = 63) | ECMO + Impella (n = 21) | ECMO (n = 42) | P-value |
|---|---|---|---|---|
| Hospital mortality, n (%) | 41 (65) | 10 (48) | 31 (74) | 0.04 |
| Bridge to next therapy or recovery, n (%) | 28 (44) | 13 (62) | 15 (36) | 0.048 |
| Weaning from MCS, n (%) | 26 (41) | 10 (48) | 16 (28) | 0.047 |
| Bridge to recovery, n (%) | 19 (30) | 8 (38) | 11 (26) | 0.3 |
| Bridge to VAD, n (%) | 8 (13) | 4 (19) | 4 (9.5) | 0.5 |
| Bridge to cardiac transplantation, n (%) | 0 | 0 | 0 | |
| Duration of ECMO, h | 120 (36–234) | 148 (72–239) | 73.5 (29–217) | 0.2 |
| Duration of MV, h | 93 (29–228) | 163 (90–228) | 48 (17–265) | 0.04 |
| CVVH, n (%) | 18 (29) | 10 (48) | 8 (19) | 0.02 |
| Haemolysis, n (%) | 30 (48) | 16 (76) | 14 (33) | 0.004 |
| Major bleeding, n (%) | 20 (32) | 8 (38) | 12 (29) | 0.6 |
| Minor bleeding, n (%) | 14 (22) | 4 (19) | 10 (24) | 0.8 |
| LVEF at weaning, % | 45.5 (30–55) | 52.5 (47–55.5) | 37.5 (25–50) | 0.13 |
- CVVH, continuous veno-venous haemofiltration; MCS, mechanical circulatory support; MV, mechanical ventilation; VAD, ventricular assist device.
Patients treated with concomitant VA-ECMO and Impella had a significantly higher rate of need for CVVH (48% vs. 19%, P = 0.02), haemolysis (76% vs. 33%, P = 0.004), and longer mechanical ventilation time [163 (90–228) h vs. 48 (17–265) h, P = 0.04] compared with VA-ECMO alone patients (Table 3).
The multiple analysis performed to identify the predictors of hospital mortality did not find covariates other than the treatment assignment.
Interestingly, patients in the study group who had myocardial recovery and weaning had a higher LVEF at discharge as compared with the control group, though this was not statistically significant (52.5 vs. 37.5%, P = 0.1).
Additional baseline and treatment characteristics
In order to minimize selection biases and treatment biases and to enhance comparability of the VA-ECMO and Impella group with the VA-ECMO group, we performed further comparative analyses with respect to baseline characteristics (Table 4) and treatment characteristics (Table 5) based on the propensity-matched sample. With regard to these clinically important parameters, no statistically relevant differences were detected between the two groups.
| Parameter | Total (n = 63) | ECMO + Impella (n = 21) | ECMO (n = 42) | P-value |
|---|---|---|---|---|
| Time from index event to ECMO (h) | 3 (1–8) | 3 (2–10) | 3 (1–8) | 0.6 |
| Cardiac rhythm at index event | 0.4 | |||
| VF | 17 (27) | 6 (29) | 11 (27) | 1 |
| VT | 1 (2) | 1 (5) | 0 (0) | 0.333 |
| Asystolia | 7 (11) | 4 (19) | 3 (7) | 0.209 |
| Sinus tachycardia/bradycardia | 30 (48) | 9 (43) | 21 (55) | 0.593 |
| AF | 6 (10) | 1 (5) | 5 (12) | 0.654 |
| PEA | 1 (2) | 0 (0) | 1 (2) | 1 |
| Out of hospital index event | 31 (50) | 12 (57) | 19 (46) | 0.7 |
- PEA, pulseless electrical activity; VF, ventricular fibrillation; VT, ventricular tachycardia.
| Parameter | Total (n = 63) | ECMO + Impella (n = 21) | ECMO (n = 42) | P-value |
|---|---|---|---|---|
| Mode of death | 0.2 | |||
| 1 Death due to heart failure | 31 (51) | 8 (38) | 23 (58) | 0.212 |
| 2 All-cause death | 5 (8) | 1 (5) | 4 (10) | 0.657 |
| Distal femoral perfusion | 39 (65) | 12 (60) | 27 (68) | 0.6 |
| Maximum inotropic score at Day 1 | 20 (15–40) | 20 (15–40) | 20 (18–40) | 0.8 |
| Maximum inotropic score at Day 2 | 11 (5–20) | 12 (10–15.5) | 10 (5–20) | 0.7 |
| Maximum inotropic score at Day 3 | 6.5 (1–15) | 9.5 (2–14.5) | 5 (0–15) | 0.7 |
| Maximum inotropic score at Day 4 | 5 (0–10) | 5 (0–10) | 5 (0–12) | 0.8 |
| Maximum inotropic score at Day 5 | 5 (0–10) | 5 (0–10) | 5 (0–8) | 0.5 |
- Inotropic score was calculated as follows: dopamine dose (µg/kg/min) × 1 + dobutamine dose (µg/kg/min) × 1 + adrenaline dose (µg/kg/min) × 100 + noradrenaline dose (µg/kg/min) × 100 + phenylephrine dose (µg/kg/min) × 100.
Inotropic score
In order to gain further information on whether the two groups were comparable in terms of treatment of CS (distribution of more favourable vs. unfavourable patients), the inotropic score was calculated over the first 5 days (Table 5). The inotropic score calculates the amount of inotropes (catecholamines and inodilators), which are administered to the patient converting the dosage in micrograms into a numerical score. None of the calculated maximum inotropic scores showed a statistical difference between the VA-ECMO group and the VA-ECMO and Impella group.
SAVE score
In order to objectivize relevant outcome parameters, we calculated the SAVE score for the propensity-matched sample (Table 6). The SAVE score has recently been validated as a reliable prognostic tool in a comparable population of CS patients under VA-ECMO treatment. The SAVE score is a survival prediction tool for refractory CS patients receiving VA-ECMO, obtained from a database of ∼3800 patients.21 No significant difference was observed at baseline between the two matched groups, confirming the comparability of the propensity-matched populations.
| Parameter | Total (n = 63) | ECMO + Impella (n = 21) | ECMO (n = 42) | P-value |
|---|---|---|---|---|
| SAVE score | −8 (−12 to −4) | −9 (−12 to −2) | −8 (−11 to −6) | 0.8 |
| SAVE score variables | ||||
| Diagnosis | 0.026 | |||
| Myocarditis | 5 (8) | 4 (19) | 1 (2) | 0.04 |
| Refractory VT/VF | 11 (17) | 1 (5) | 10 (24) | 0.08 |
| Post-HTX or LTX | 1 (2) | 0 (0) | 1 (2) | 1 |
| CHD | 0 (0) | 0 (0) | 0 (0) | N/A |
| Other | 46 (73) | 16 (76) | 30 (71) | 0.8 |
| Age (years) | 0.3 | |||
| 18–38 | 6 (10) | 4 (19) | 2 (5) | 0.089 |
| 39–52 | 20 (32) | 7 (33) | 13 (31) | 0.848 |
| 53–62 | 16 (25) | 4 (19) | 12 (29) | 0.544 |
| >62 | 21 (33) | 6 (29) | 15 (36) | 0.571 |
| Weight (kg) | 0.2 | |||
| <65 | 5 (8) | 1 (5) | 4 (10) | 0.657 |
| 65–89 | 44 (71) | 18 (86) | 26 (63) | 0.080 |
| >89 | 13 (21) | 2 (10) | 11 (27) | 0.189 |
| Acute pre-ECMO organ failures | ||||
| Liver failure | 26 (41) | 9 (43) | 17 (40) | 0.9 |
| CNS dysfunction | 24 (38) | 10 (48) | 14 (33) | 0.3 |
| Renal failure | 38 (60) | 14 (67) | 24 (57) | 0.5 |
| Chronic renal failure | 14 (22) | 3 (14) | 11 (26) | 0.4 |
| Duration of intubation prior to initiation of ECMO (h) | 0.7 | |||
| <11 | 55 (86) | 18 (86) | 37 (87) | 1 |
| 11–29 | 6 (10) | 3 (14) | 3 (7) | 0.391 |
| >29 | 2 (3) | 0 (0) | 2 (5) | 0.548 |
| Peak inspiratory pressure ≤20 cmH2O | 11 (17) | 4 (19) | 7 (17) | 1 |
| Cardiac arrest before ECMO | 37 (59) | 12 (57) | 25 (60) | 0.5 |
| Diastolic BP before ECMO ≥40 mmHg | 44 (70) | 17 (81) | 27 (64) | 0.2 |
| Pulse pressure before ECMO ≤20 mmHg | 35 (56) | 13 (62) | 22 (52) | 0.5 |
| HCO3 before ECMO ≤15 mmol/L | 19 (30) | 6 (29) | 13 (31) | 0.8 |
- The SAVE score reveals no significant difference between the study group and the control group in the propensity-matched cohort.
- For the heading ‘Acute pre-ECMO organ failure’ there was no P-value calculated, because some patients experienced more than one acute pre-ECMO organ failure (e.g. kidney and liver failure) at the same time. Therefore, the calculation of a P-value for this heading would be statistically misleading.
- BP, blood pressure; CHD, chronic heart disease; CNS, central nervous system; HTX, heart transplantation; LTX, lung transplantation; N/A, not applicable; SAVE, survival after veno-arterial ECMO; VT, ventricular tachycardia; VF, ventricular fibrillation.
Discussion
The important finding of this study is that combined CS treatment with Impella and VA-ECMO proved to be safe and feasible. It is further associated with improved outcome in this large retrospective, multicentre, propensity-matched study of patients presenting with CS compared with patients treated with VA-ECMO alone. This represents a further step in improving the results of this challenging clinical setting.
The use of VA-ECMO is an established therapy option in severe CS, since femoro-femoral VA-ECMO can be implanted rapidly and improves the haemodynamic situation immediately.3 Nevertheless, mortality of CS despite VA-ECMO is still high.24 Importantly, no large controlled trial has so far addressed the application of VA-ECMO therapy.
Lack of LV unloading is one of the unsolved problems associated with serious complications and reduced outcome which limits the applicability of VA-ECMO.7 Indeed, highly increased afterload is the haemodynamic picture of VA-ECMO itself. Importantly, lowering afterload is one of the main targets of any successful heart failure medical or device therapy. Moreover, any complications such as LV stasis with thrombus formation11, 25 as well as pulmonary failure due to high pulmonary artery pressures9, 10, 25 and myocardial ischaemia caused by ventricular distension8 are complications of VA-ECMO treatment. Therefore, therapeutic options for LV unloading might be beneficial for the patients treated with VA-ECMO.26 This was shown in a small case series where an IABP was combined with VA-ECMO.15 In that study, 135 patients were treated, and the authors showed that, with an in-hospital mortality rate of 42.2% and complications such as bleeding, stroke, and vascular complications between 11% and 16%, the additional use of IABP on top of VA-ECMO is a reasonable treatment option. This indicates the potential success of mechanically reducing afterload in VA-ECMO-treated patients. Unfortunately, a comparison with VA-ECMO-only patients was not possible, since no controls were presented in that study.
Other approaches such as cannulation at the LA13, 27 or the LV7 were described, but no studies evaluated these approaches in a more detailed way. Apart from this, the cannulation methods of LV/LA venting are surgical interventions with high risk of serious complications. The Impella device, in contrast, can be inserted percutaneously, avoiding the need for surgical interventions, and can provide a higher degree of haemodynamic support compared with an IABP. Our study utilized a multicentric, retrospective cohort of 157 patients with profound refractory CS treated with VA-ECMO or the combination of Impella and VA-ECMO to improve LV unloading. It is the largest cohort ever investigated in this view. In our study, patients treated with concomitant VA-ECMO and Impella have a lower hospital mortality and a higher rate of successful bridging to either recovery or next therapy as compared with patients treated with VA-ECMO alone.
Since we documented differences in baseline characteristics, especially with respect to potentially critical prognostic parameters such as pre-ECMO cardiac arrest,21 higher percentage of CPR, lower pH, and a trend to higher lactate, we performed a propensity matching in which these variables amongst others were evenly distributed between the VA-ECMO group and the VA-ECMO and Impella group. After this, we could still show improved outcome and increased bridging to recovery or next therapy in the combined approach. We extended the analyses and complemented the propensity-matched cohort by adding further baseline as well as treatment variables that might have an impact on outcome. There were no statistically significant differences between the VA-ECMO and Impella group and the VA-ECMO group, corroborating our initial results. Additionally, we calculated the SAVE score as an objective means of risk estimation in VA-EMCO patients in order to minimize selection biases and treatment biases further. Again, there was no difference observed between the propensity-matched cohorts. Thus, considering the retrospective nature of the study, we were able to match cases and controls and were able to minimize confounding factors that may otherwise have biased the results.
Notably, myocarditis incidence was higher in the VA-ECMO and Impella group (4 vs. 1 case, P = 0.04) compared with controls, suggesting a greater potential for recovery in the treatment group. On the other hand, the number of cases is extremely low and it seems speculative to argue an effect on the outcome in our specific population. Furthermore, this variable was included in the SAVE score, which strongly supports an even distribution of outcome-related patients' characteristics. Thus, we are convinced that the higher incidence of myocarditis in the VA-ECMO and Impella group does not depict a relevant bias in the interpretation of the results.
A few of the patients in this study were initially treated with an IABP for a short period of time. In our analyses, we did not address the role of IABP in CS, but its concomitant application during CS treatment was taken into consideration in our analysis, being included in the performance of the propensity matching. No statistical difference was observed concerning the IABP use between the two matched populations (P = 0.3); therefore, we can argue that this variable does not affect our results.
A known limitation of the Impella device is represented by haemolysis.28-30 This observation was confirmed in our population, as we report a higher incidence of haemolysis in patients treated with VA-ECMO and Impella as compared with VA-ECMO alone. Bleeding complications did not differ and therefore had no impact on outcome in this study.
One beneficial effect of unloading the left ventricle under VA-ECMO therapy is to reduce pulmonary oedema, yet the mechanical ventilation time in the VA-ECMO and Impella group was considerably longer than in the VA-ECMO-only group. Considering this observation, one could assume that prolonged ventilation time indicates augmented manifestation of pulmonary oedema. In our opinion, though, the longer ventilation time in the VA-ECMO and Impella group reflects the improved survival of this group. Obviously, the additional utilization of Impella on top of VA-ECMO does not lead to a quicker recovery, but to a recovery of patients in severe CS who might not have survived under VA-ECMO-only therapy. This observation goes along with the longer ECMO time in the VA-ECMO and Impella group, and should be similarly interpreted. Reinforcing our opinion is a recent large-scale database evaluation of patients treated with VA-ECMO for CS, which suggests that those patients who survived CS and were discharged from hospital had a significantly longer duration of VA-ECMO support compared with those patients who died from CS.21
Limitations
One limitation of our study concerns its retrospective nature, with its natural lack of a prospective design. However, this clinical attitude was independently observed in two different centres, which had contemporarily documented encouraging clinical results.
Another limitation is represented by a potential bias in the selection of patients described here, especially since the decision for implantation of either VA-ECMO or VA-ECMO and Impella was deemed appropriate by the attending physician in the absence of a specific protocol. On the other hand, the results of this study are based on extended statistical means to maximize comparability of the study group and the control group and on application of the latest VA-ECMO-related risk score analyses (SAVE score). Furthermore, the decision to implant the Impella was driven by the evidence of the need for LV venting (stone heart, massive pulmonary oedema, impending clotting on the left ventricle, significant aortic regurgitation) at the time of VA-ECMO implantation, that would also require that patients in the study group were sicker at baseline, which further reinforces our results. Taken together, selection biases are possible and need to be kept in mind when interpreting the results of our study.
The proposed approach possibly changes the weaning process as well, as it might hasten removal from VA-ECMO as the left ventricle is still partially supported. This might translate into a lower rate of VA-ECMO-related complications and increased myocardial recovery through prolonged unloading. However, this hypothesis is only speculative at this stage and warrants further investigation in larger studies.
Another limitation is the lack of mechanistic insight of this study, which is explained by the fact that data were documented in clinical routine. In addition, we are fully aware that some relevant clinical variables, including the high rate of CPR in both groups, may have a major influence on the patients' profile and outcome despite a good propensity matching, but these data reflect the real-world clinical scenario of the treatment of CS.
Conclusion
We observed significantly improved survival of patients with life-threatening CS and VA-ECMO-based mechanical circulatory support by additional implantation of Impella. These results indicate a potential survival benefit of the combinatory approach utilizing VA-ECMO and Impella. Considering the retrospective nature of this study, future prospective, randomized studies are needed to validate these results and possibly contribute to improvements in the acute management of severe CS.
In our experience, combined support with VA-ECMO and Impella is associated with reduced hospital mortality and a higher rate of successful bridging to either recovery or next therapy (LVAD implantation or heart transplantation) in patients with refractory CS. These results may be explained by the effectiveness of Impella in relieving VA-ECMO-related LV overload and its consequences (subendocardial malperfusion, pulmonary venous congestion, and LV stasis) and by the increased net forward flow. Future randomiszed studies are warranted to validate this strategy.
Funding
The study was supported by departmental funds.
Conflict of interest: none declared.
