Introduction

Cardiogenic shock is the inability of the heart to provide adequate blood flow to the organs. It

is defined by the combination of a systolic blood pressure of less than 90 mmHg, a cardiac

index of less than 2.2L/min/m², and a Pulmonary capillary wedge pressure (PCWP)) greater

than 15 mmHg. It is qualified as refractory when acute circulatory insufficiency persists

despite optimization of maximum conventional therapies and then justifies the implantation of

mechanical circulatory assistance.

ECLS is used as a substitute for left, right or global cardiac dysfunction and is also equipped

with an oxygenator. It improves the survival of patients (5,6)in whom it is used, waiting for

myocardial recovery, as a bridge to heart transplantation or implantation of a long term

ventricular assistance.

However, its benefits may be outweighed by changes in loading conditions, particularly in the

case of peripheral cannulation of the device where blood is re-injected countercurrently into

the arterial circulation, resulting in an increase in left ventricular (LV) afterload. (7). These

changes in loading conditions could impair myocardial recovery. (BIBLIO).

The IMPELLA® is a continuous flow axial micro-pump that is endovascularly placed in the

transvalvular aortic position, drawing blood from the LV and reinjecting it into the ascending

aorta. It then acts as a true discharge of the LV, reducing its afterload, while improving

coronary perfusion.

Some teams have recently proposed to combine ECLS and IMPELLA® in patients in

refractory cardiogenic shock with severe left heart dysfunction, the objective being to

generate a high level of high-flow circulatory assistance through the ECLS, while ensuring

effective left ventricular discharge through the IMPELLA®.

Several studies have shown conflicting survival results [PATEL and PAPPALARDO;

Circulatory Support with ECLS and/or Impella for cardiogenic shock during MI, MOURAD,

GAUDARD] in patients receiving ECLS and IMPELLA® in combination versus ECLS

alone.

In the light of these divergent results found in the literature, we have carried out a study aimed

mainly at studying the effect of the combination of ECLS and IMPELLA® taking into

account the temporality of implantation.

Materials and methods

Study design

This retrospective, monocentric study was conducted in the Acquired Cardiopathies Unit of

the Southern Anesthesia-Resuscitation Department of the Bordeaux Teaching Hospital

(France), between January 2011 and June 2018. It was the subject of a declaration to the Data

Protection Officer of the Bordeaux Teaching Hospital (France).

Patients

Patients over 18 years of age with refractory cardiogenic shock who received ECLS alone or

ECLS in combination with IMPELLA® were included in this study. Cardiogenic shock was

defined as a combination of systolic blood pressure less than 90 mmHg, cardiac index less

than 2L/min/m², PAPO greater than 15 mmHg, despite correct blood volume, inotropic peak

dose treatment (dobutamine, adrenaline and/or milrinone respectively at a dose higher than 10

µg/kg/min, 0.1 µg/kg/min, 0.5 µg/kg/min), a venous oxygen saturation (SvO2), taken from

the pulmonary artery catheter, of less than 55% with a haemoglobin higher than 10g/dL.

The inclusion criteria were: circulatory support by ECLS alone or ECLS in combination with

IMPELLA®, 2.5 or CP or 5.0, whose indication was cardiogenic shock secondary to

myocardial infarction, decompensation of ischemic, dilated, rhythmic, valvular, hypertrophic

or mixed heart disease, or acute myocarditis.

The exclusion criteria were: circulatory support by IMPELLA® alone, support indicated for

post-cardiotomy, or for cardiotropic intoxication, or for severe pulmonary embolism, or other

rare indications (tumour compression, hypothermia).

Implementation of circulatory assistance

The decision to assist a patient in refractory cardiogenic shock was made following a

consultation between cardiac surgeon, cardiologist, and anesthesiologist, during which the

patient's co-morbidities and the context in which the state of shock occurred were taken into

consideration. This "SHOCK-TEAM" also took into account the characteristics of the failures

to be replaced (degree of left and/or right cardiac dysfunction, degree of hypoxemia) to guide

the choice of the device to be implanted.

 Choc cardiogénique réfractaire en l’absence de Algorithme hôpital cardiologique
contre-indications à l’ACT CHU de Bordeaux (Janvier 2013)

- Urgence vitale immédiate (ACR ou orage rythmique)

- Dysfonction biventriculaire sévère
- Pronostic neurologique incertain
- Contre-indication à l’IMPELLA (thrombus intra-VG…)

OUI ECMO VA
et
NON Réévaluation secondaire

- IMPELLA 5.0 ou CP (abord axillaire+++) sous contrôle ETO

- Nouvelle évaluation hémodynamique à 30 min:
• ETO (FE, DTDVG, TAPSE, onde S’)
• Swan-Ganz (ScvO2, IC, PAPO, PAP, PAPi, POD, CPO)
• Gazométrie artérielle (rapport PaO²/FiO²)

Absence d’hypoxémie Hypoxémie sévère Absence d’hypoxémie Hypoxémie sévère

et mais mais et
Fonction VD conservée Fonction VD conservée Fonction VD altérée Fonction VD altérée

ECMO OD-AP ECMO OD-AP

IMPELLA seule ECMO VV
ou ou
ECMO VA ECLS hybride

Figure 1: MCS implantation algorithm (Bordeaux Teaching Hospital)

The ECLS could be placed peripherally, with in the majority of cases a femoral-femoral

cannulation, more rarely a femoro-axillary or central cannulation.

The increase in LV afterload was defined by the occurrence of clinical and/or radiological

signs of acute lung oedema on the one hand, and, on the other hand, on echocardiography, by

the observation of LV dilatation accompanied by functional mitral insufficiency or even

sludge in the left cavities, or an elevation of the OAP.

All three IMPELLA® models, 2.5, CP and 5.0 were used. The 2.5 and CP pumps could be

implanted percutaneously through the femur or surgically through the axilla. The 5.0 pump

was implanted surgically at the axillary level only.

The "ECMELLA" group consisted of patients assisted by the combination of ECLS and

IMPELLA®, the "ECLS Alone" group by patients assisted by ECLS alone.

Patients remaining on IMPELLA® alone were excluded from the analysis.

The combination of an IMPELLA® and an ECLS could be simultaneous ("Simultaneous

ECMELLA") or delayed ("Sequential ECMELLA").

Patient management under circulatory support

The patients included were all hospitalized in our adult cardiac intensive care unit.

Adjustment of the level of circulatory support (ECLS and IMPELLA®) and the dosage of

inotropic and vasopressor drugs was guided by the hemodynamic evaluation of the patients,

assessed on clinical examination, analysis of scope parameters (Electrocardiogram,

monitoring of systemic and pulmonary pressures (Swan-Ganz catheter), monitoring of

peripheral oxygen saturation (SpO2)), on biological tests (arterial and venous gasometry

allowing in particular the monitoring of blood sugar and SvO2, renal and hepatic assessment

and BNP (Brain natriuretic peptide) assay), and on the results of echocardiography

(evaluation of myocardial recovery, search for dilatation of the cardiac cavities, search for

thrombus, study of the functionality of the valves, verification of the position of the assistance

devices (ECLS admission cannula, position of the IMPELLA® catheter). Oxygenation

adjustment was based on clinical assessment, SpO2 monitoring by positioning the sensor on

the right upper limb, and PaO2 analysis on arterial gasometry, by placing the sampling site

also on the right upper limb, in order to detect early on cerebral hypoxemia that may

potentially occur in patients with damaged lung with effective ejection through the aortic

valve by myocardial recovery or IMPELLA® placement. This monitoring was supplemented

in sedated patients by monitoring of cerebral Near-infrared spectroscopy (NIRS).

Weaning
The therapeutic orientation of each patient (withdrawal from ACT, transplantation,

implantation of long-term left mono-ventricular assistance) was the subject of regular

multidisciplinary discussion, and was established taking into account age, comorbidities and

social environment, clinical history, hemodynamic evolution in resuscitation, and the

occurrence of possible intercurrent complications.

ECLS and IMPELLA® could be withdrawn at the same time or delayed over several days,

either one first. In our analysis, we do not differentiate according to the sequence of

explantation. In patients on ECLS alone or ECMELLA progressing to refractory isolated left

ventricular dysfunction, a switch to assistance with IMPELLA® 5.0 alone was made

whenever possible. Patients on ECLS alone were considered weaned from ACT when they

had a stable haemodynamic state without haematosis disorder and without clinico-biological

evidence of peripheral hypoperfusion under low levels of assistance (flow rate less than 2.5

L/min), low doses of inotropes and vasopressors (dobutamine < 2.5 gamma/kg/min, milrinone

< 0.2 gamma/kg/min, norepinephrine < 0.1 gamma/kg/min, epinephrine < 0.1

gamma/kg/min), and evidence for myocardial recovery on echocardiography (subartic VIT >

10 cm) and pulmonary artery catheter evaluation (Cardiac Index > 2.5 L/min/m², PAPO < 15

mmHg, SvO2 > 65%).

Patients on IMPELLA® alone were considered to be weaned from assistance when they had a

stabilized hemodynamic state at rest or even during exercise (walking, cycling) at reduced

pump performance at P2 for more than 5 days, and there was evidence of myocardial recovery

on echocardiography (subartic VTBI > 10 cm) and pulmonary artery catheter evaluation

(Cardiac Index > 2.5 L/min/m², PAPO < 15 mmHg, SvO2 > 65%).

Adjuvant Therapies
The vast majority of patients benefited from inotropic and vasopressor support prior to the

installation of circulatory support. Some of these treatments were continued after

implantation. The drugs used were norepinephrine, dobutamine, milrinone and adrenaline. For

associated right heart failure, some patients received inhaled nitric oxide.

All ECLS-only subjects were curatively anticoagulated with unfractionated heparin using an

electric syringe. Those on ECMELLA received, in addition to this therapy, anticoagulation of

IMPELLA® via the purge system with unfractionated heparin diluted in glucose serum.

Anticoagulation was monitored by antiXa activity, with targets ranging from 0.3 to 0.4

IU/mL.

Data collection

The data collected for the period prior to the implantation of the support were as follows:

general characteristics of the patients, indication for the support, occurrence of cardiovascular

arrest prior to the implantation of the support, type of support, troponin, BNP, lactates,

transaminases, bilirubin, prothrombin level.

Judgement criteria

The primary endpoint was resuscitation mortality.

Secondary endpoints were complications, length of assistance and stay, patient outcome

(withdrawal from assistance, bridge-to-LVAD (Left Ventricular Assist Device), bridge-to-

transplantation), 6-month survival, use of extra-renal lavage, vasopressor and inotropic drugs,

and changes in biological parameters.

Statistical analysis
Qualitative variables were expressed in numbers and percentages. Their comparison was

made with the Fisher exact test or the Chi² test, depending on the application conditions.

Continuous quantitative variables were expressed as the median associated with their

interquartile range. They were compared with a Student's test or a non-parametric Wilcoxon-

Mann-Whitney test, depending on whether the variable being compared had a normal

distribution or not. Comparisons of the continuous quantitative variables according to 3

groups were carried out with the non-parametric Kruskal-Wallis test.

Comparisons were first made between patients on ECLS alone versus patients on ECMELLA.

If there was a significant difference in the overall test, a post-hoc analysis was performed

using the ECLS alone group as a reference. In this case, the inflation of the alpha risk was

taken into account and the p-values corrected with the Hochberg correction method.

The normality of the distributions was evaluated graphically with histograms and confirmed

statistically with the Kolmogorov-Smirnov test.

Statistical significance was defined by a p-value less than or equal to 0.05. The analyses were

performed with R software version 3.5.1.

Results

Of the 332 patients in refractory cardiogenic shock treated with ACT between January 2011

and June 2018 in our intensive care unit, 140 were included in the study. Eighty-two patients

(58.6%) were treated with ECLS alone (ECLS Alone group), and 58 patients (41.4%) were

treated with ECLS plus IMPELLA® combination (ECMELLA group), with 30 subjects (21%

of included patients) implanted as dual-assisted patients from the outset, and 28 subjects (20%

of included patients) implanted sequentially (Figure 1).

***

The characteristics of the general population and the ECLS Alone and ECMELLA groups (all

association sequences combined) are detailed in Table 1. The 2 populations were comparable
for median age (54 years [49; 62]), anthropometric data and comorbidities except for the

history of vascular pathology (stroke, obliterative arterial disease of the lower limbs (AOMI)),

which was significantly more frequent in the ECLS Alone group (57.3% versus 10.5%,

p<0.001). There was also a significantly higher proportion of males in this sample (88%

versus 69%, p=0.01). The "ischemic" indication for ACT was significantly more frequent in

the ECMELLA group (65.5% versus 40.2%, p=0.001), with a parallel increase in troponin

(242 [IQR 31; 421.3] versus 6.6 [IQR 0.9; 48], p<0.001). The median SAPSII severity score

was 66 in both groups. Finally, there was a higher proportion of patients who experienced

cardiocirculatory arrest prior to the implementation of ACT in the ECLS Alone group (57.3%

versus 41.4%, p=0.063).

The characteristics of ACT are described in Table 2. Of the total study population, 131

patients (93.6% of cases) were implanted with femoral ECLS.

In the ECMELLA group, IMPELLA® 5.0 was the most widely implemented model (n=39, or

67.2% of cases). In the Simultaneous ECMELLA group, IMPELLA® 2.5, CP and 5.0 were

implanted in 30%, 10% and 60% of patients respectively. In the Sequential ECMELLA

group, IMPELLA® 2.5, CP and 5.0 were implanted in 14%, 11% and 75% of patients,

respectively.

ECLS hybridization was performed in 15 patients (10.7% of cases) and significantly more

frequently in the ECMELLA group than in the ECLS Alone group (respectively 11 patients,

or 18.9% of cases versus 4 patients, or 4.9% of cases, p=0.01).

The median duration of assistance per was 5 days [3; 8] on ECLS alone, and 7 days [3; 13.8]

on IMPELLA® alone. The median duration of the combination of the two treatments was 4.5

days [1; 7]. There was no significant difference between the ECLS Alone and ECMELLA

groups.
In contrast, the median value of the total duration of assistance (ECLS +/- IMPELLA® alone

or in combination) was 7 days [4; 14] in the general population and was significantly lower in

the ECLS Alone group (5 [3; 7] versus 14 [8; 20], p<0.001).

Analysis of the frequency of thromboembolic and hemorrhagic complications did not reveal

any significant differences between the two ECLS Alone and ECMELLA groups (Table 3) or

between the 3 ECLS Alone, Simultaneous ECMELLA, and Sequential ECMELLA groups

(Table 4).

Of the total study population, 10 patients had an ischemic stroke, and 14 had other arterial

ischemic complications (acute limb ischemia, mesenteric ischemia). IMPELLA® thrombosis

occurred in 11 patients (5.2% of cases) and ECLS thrombosis in 3 patients (7.9% of cases),

with no significant difference between the groups.

Thirty-five patients in the ECLS Alone group (42.7%) developed bleeding requiring

transfusion during their resuscitation stay, compared with 23 in the ECMELLA group

(39.7%) (p=0.720). Twenty-two patients in the ECLS Alone group (26.8% of cases) were

taken back to the OR, mainly for surgical haemostasis, compared with 14 patients in the

ECMELLA group (24.1% of cases) (p=0.720).

Patient outcomes after ACT implantation are described in Table 3. ACT was successfully

weaned (weaned from all non-transplant assistance for at least 7 days) in 47 patients (57.3%)

in the ECLS Alone group versus 23 (39.7%) in the ECMELLA group, (p=0.04).

Of the 140 patients included in the analysis, 88 (62.9%) were resuscitated alive, of which 63

(45%) were weaned from any ACT, 18 (12.9%) were implanted with long-term left mono-

ventricular assistance and 7 (5%) were transplanted. Mortality in resuscitation was

comparable between the ECLS Alone and ECMELLA groups (respectively 35.4% and

39.7% , p=0.605). It should be noted that a higher proportion of patients in the ECLS Alone
group were discharged from resuscitation without any assistance (53.7% versus 32.8,

p=0.014), while a higher proportion of patients in the ECMELLA group were discharged

from implanted resuscitation with left mono-ventricular assistance (6.1% versus 22.1%,

p=0.005). There was no significant difference in live resuscitation output after transplantation

(respectively 4 patients (4.9%) in the ECLS Alone group versus 3 (5.2%) in the ECMELLA

group).

For the analysis of the temporality of the association, the median troponin value differed

significantly between patients in the ECLS Alone, Simultaneous ECMELLA, and Sequential

ECMELLA groups (respectively, 6.61 [0.9; 48] versus 158.5 [20]).3 ; 331] versus 311 [37 ;

590.5], p<0.001), as well as the indication of assistance for myocardial infarction

(respectively, 40.2% versus 66.7% versus 64.3% with p=0.013) and decompensation of

chronic heart disease (respectively, 52.4% versus 26.7% versus 28.6%, with p=0.013). Post-

hoc analysis showed a significant difference in troponin value, higher in the Simultaneous

ECMELLA group than in the ECLS Alone group and higher in the Sequential ECMELLA

group than in the ECLS Alone group, but did not find a significant difference in the indication

for assistance.

Further analysis of the three ECLS groups Alone, Simultaneous ECMELLA and Sequential

ECMELLA showed a significant difference in hybridization (respectively 4.9% versus 26.7%,

versus 10.7%, p=0.001), and the total duration of assistance (respectively 5 days [3; 7] versus

14 days [8; 20] versus 15 days [8; 20], p<0.001). The post-hoc analysis also found a

significant difference in the duration of any assistance, shorter in the ECLS Alone group than

in the Simultaneous ECMELLA group, and shorter in the ECLS Alone group than in the

Sequential ECMELLA group, and a hybridization more frequently performed in the

Simultaneous ECMELLA group than in the ECLS Alone group (Table 4).
In the Sequential ECMELLA group, again compared to the Simultaneous ECMELLA and

ECLS Alone groups, a lower proportion of patients were weaned from any ACT (21.4%

versus 43.3% and 53.7% respectively, p=0.012), with a post-hoc analysis showing a

significant difference between the ECLS Alone and Sequential ECMELLA groups. A greater

proportion of patients were implanted with long-term left mono-ventricular support (32.1%

versus 13.3% and 6.1% respectively, p=0.002), with a post-hoc analysis showing a significant

difference between the ECLS Alone and Sequential ECMELLA groups.

In the Sequential ECMELLA group, ECLS was the ACT initially implemented in 71.4% of

cases.
Discussion

Our study found no difference in mortality between the ECLS Alone and ECMELLA groups,

both at the resuscitation exit (respectively 35.4% versus 39.7%, p=0.605) and at 6 months

(39.1% versus 50.9%, p=0.169). These results were not influenced by the temporality of the

combination of ECLS and IMPELLA® : there was no significant difference in mortality

between the ECLS Alone, Simultanenous ECMELLA and Sequential ECMELLA groups.

In the ECMELLA group, patients were, compared to the ECLS Alone group, less frequently

weaned from resuscitation (19 cases (32.8%) versus 44 (53.7%), p=0.014) and more

frequently implanted with long-term left mono-ventricular assistance (13 cases (22.1%)

versus 6 cases (5.1%), p=0.005). This difference is not found in the literature (BIBLIO

PATEL+PAPPALARDO).

Temporality appeared to influence the outcome of patients discharged alive from resuscitation

(Table 4). Patients in the ECLS Alone and Simultaneous ECMELLA groups were more

frequently weaned and alive at the resuscitation exit than patients in the Sequential

ECMELLA group (44 patients (53.7%) versus 13 (43.3%) versus 6 (21.4%), p=0.012) and

less frequently implanted with a LVAD at the resuscitation exit (4 patients (13.3%) versus 4

(13.3%) versus 9 (32.1%), p=0.002). Post-hoc analysis confirmed these differences between

the ECLS Alone and Sequential ECMELLA groups. The question of patient acuity between

these three groups deserves to be asked. However, there were no significant differences in

terms of SAPSII score, pre-implantation lactate, occurrence of pre-implantation RCA, or

indication for ACT (Table 4). The Sequential ECMELLA group probably consisted of

patients with more severe or secondarily worsening conditions that warranted the implantation

of an IMPELLA® in addition to ECLS to offload a failing left ventricle and whose increased
afterload on ECLS was poorly tolerated. The more pejorative outcome of patients in the

Sequential ECMELLA group raises the question of the timing of the IMPELLA®

implantation in addition to ECLS: was it too late, too early and with complications, or were

the patients initially more severe?

The Patel and Pappalardo studies report higher hospital discharge mortality figures, and a

significant difference between the two groups with a death rate of 57% and 78% in the ECLS

alone and ECMELLA groups respectively in the first study and a death rate of 48% and 74%

in the second study. These discrepancies with the literature can certainly be explained by the

small numbers studied and the variable medical practices according to the teams, particularly

with regard to the strategy for the indication and management of ACTs. Thus, in Pappalardo's

study, 19% of patients were assisted by ECMELLA for acute myocarditis, compared to 7% in

our study.

The absence of a significant difference in post-resuscitation mortality between our two groups

should certainly not lead to the conclusion that IMPELLA® in combination with ECLS is of

no value. On the contrary, the combination of IMPELLA® and ECLS to relieve LV improves

the survival of patients in this group, most of whom are being treated for myocardial

infarction and who have a definite benefit to be gained from relieving an ischaemic left

ventricle.

The characteristics of our population (median age of 54 years, 80% male) are similar to those

of similar studies. (BIBLIO Predicting survival after ECLS for refractory cardiogenic shock: the survival after veno-arterial-ECLS (SAVE)-score).

In our work, ACT has been implanted in 50.7% of cases in patients with refractory

cardiogenic shock complicating myocardial infarction, which is consistent with the series of

Pappalardo et al. where temporary circulatory support in this context is found in 48% of cases.

In other series encompassing a larger patient population, managed for medically induced

cardiogenic shock or occurring post-cardiotomy, the indication for ACT for acute coronary
syndrome with refractory shock represents less than one third of cases. ( ESLO cohort, Patel, R. García-
Gigorro, Graham Peigh, Cécile Aubron, Michael M. Koerner).

With a median score of 66 [51; 83], the SAPSII score of our population was higher than in

other studies reporting a score close to 50-60 (BIBLIO García-Gigorro R, Renes-Carreno E, Pérez-Velaa JL et al. Mechanical

support with venoarterial extracorporeal membrane oxygenation (ECLS-VA): sPeigh G, Cavarocchi N, Keith SW et al. Simple new risk score model for adult

cardiac sextracorporeal membrane oxygenation. J Surg Res. 2015 Oct;198(2):273-9).

The maximum value of troponin prior to ACT implantation was significantly higher in the

ECMELLA group than in the ECLS Alone group (242 ng/mL [31; 421.3] versus 6.6 ng/mL

[0.9; 48], p<0.001, respectively). This may be explained by the majority indication of ACT

for myocardial infarction in the ECMELLA group. Three-group analysis reveals, probably for

the same reason, that the maximum pre-ACT troponin is significantly higher in the

Simultaneous ECMELLA and Sequential ECMELLA groups than in the ECLS Alone group,

with post-hoc analysis confirming these differences.

In our cohort, the results observed on the lactate level before the implantation of the

assistance (BIBLIO PATEL AND PAPPALARDO) and the frequency of pre-ACT cardio-

circulatory arrest (50.7% of cases) are close to those from other studies (PAPPALARDO

BIBLIO ), whose population is comparable, but more distant from those of other studies

(Garcia-Gigorro et al.) that included in the analysis patients assisted post-cardiotomy, who

reported lower lactate levels and a lower proportion of patients with cardiovascular arrest

prior to the implantation of ACT.

In our population, the indication for ACT for myocardial infarction was significantly more

frequent in the ECMELLA group than in the ECLS Alone group (respectively 38 patients

(65.5%) versus 33 (40.2%), p=0.001). This difference may be explained by the high

frequency of damage to the left coronary network in myocardial infarction resulting in

refractory cardiogenic shock, where severe left heart dysfunction most often justifies the use

of IMPELLA® in combination with ECLS to discharge LV.

In our work, the median value of the total duration of ACT in the ECLS Alone group was

close to that reported in the literature (BIBLIO Cécile Aubron, Predicting survival after ECLS for refractory cardiogenic shock: the
survival after veno-arterial-ECLS (SAVE)-score Matthieu Schmidt1, Duration of veno-arterial extracorporeal life support (VA ECLS) and outcome: an

analysis of the Extracorporeal Life Support Organization (ELSO) registry Myles Smith, ETUDE PATEL,). It was also significantly lower

than in the ECMELLA group (5 days [3; 7] versus 14 days [8; 20], p<0.001). It should be

noted here that no comparative data on the duration of assistance under ECLS alone and

ECMELLA appear in similar publications. This difference can probably be explained by the

management strategy adopted in our department for doubly-assisted patients, which consists

of weaning them from ECLS as a priority and engaging them over several days in a

rehabilitation phase on IMPELLA® alone. Our objective is to allow time for the ventricular

remodeling phase under conditions of optimal LV discharge for better preparation for

weaning from assistance in patients whose myocardium is recovering, and to test the

eligibility for implantation of long-term left mono-ventricular assistance in those left with

high levels of IMPELLA® performance due to persistent severe LV dysfunction. Taking into

account the strategy for the indication of assistance in our department, patients in the

ECMELLA group, for whom the combination of IMPELLA® and ECLS was necessary, are

subjects whose left ventricular function was probably more severely impaired than that of

patients in whom such a combination of assistance was not retained. It is therefore consistent

to observe that in the ECMELLA group, patients in the ECMELLA group were less

frequently weaned from resuscitation and more frequently implanted with long-term left

mono-ventricular support than in the ECLS Alone group. For the ECLS Alone, Simultaneous

ECMELLA and Sequential ECMELLA groups, the median length of stay was 14 days [6; 29],

16 days [8; 30], 25 days [13; 56], p=0.06. Because there was no statistically significant

difference, post-hoc analysis was not performed. Nevertheless, the small size of the

ECMELLA group and its two subgroups may contribute to a lack of power. The length of

resuscitation stay appeared to be longer in the Sequential ECMELLA group than in the
Simultaneous ECMELLA group. This difference could firstly be explained by the sequential

nature of the implementation of the assistance.

Concerning complications that may be related to ACT, 41.4% of all patients developed a

major haemorrhage requiring transfusion, a result comparable to the meta-analysis of Cheng

et al (BIBLIO), which included 20 studies and 1,866 patients, of whom 40.8% had suffered a

major haemorrhage.

In our entire cohort, 25.7% of patients were returned to the operating room, mainly for control

of major bleeding, compared to 41.9% in the Cheng et al. study (BIBLIO). In our study, 10%

of patients presented with acute peripheral ischemia (upper or lower limb, digestive ischemia)

versus 16.9% in the Cheng et al. study (BIBLIO) (lower limb ischemia only). This difference

could be explained by different monitoring and infusion techniques for the lower limbs in

different teams. 7.9% of our patients developed an ischemic stroke, a relatively similar result

to the study by Cheng et al. (BIBLIO) where 5.9% of patients developed this complication.

Our study had several positive points. Firstly, it was large (140 patients), given the rarity of

refractory cardiogenic shock. Secondly, the indications, although only medical, were

relatively diverse, and the patients were comparable in terms of initial severity (cardio-

circulatory arrest before implantation, lactates, SAPSII score). We also had little missing data,

with only 2% lost to follow-up at 6 months.

However, we must point out some limitations regarding the robustness of our work. First, it

was a monocentric with a retrospective character. These characteristics could induce a

collection bias but also a confusion bias, as we had only included patients who were

hospitalized in the same department, cared for by one and the same medical team. The

number of staff, although important in this pathology, is nevertheless low from a statistical

point of view. A lack of power could result. Then, the evaluation of the initial seriousness of
the patients by the SAPSII score, the lactates, the occurrence of a pre-implantation cardio-

circulatory arrest could be discussed. The "intrinsic" severity of patients was probably best

represented by the indication for assistance. These indications, quite diverse, could also

induce a selection bias. Finally, we were unable to collect reliable data on ventilation, even

though the effect of a left discharge could influence it.