Post‐Market Analysis of the Xenios Console With the Xlung Tubing Kit 230 for Extracorporeal Membrane Oxygenation After Cardiac Surgery

ABSTRACT

Objectives

The use of veno‐arterial extracorporeal membrane oxygenation (V‐A ECMO) has led to an increasing number of devices available on the market. Most studies pay less attention to which devices have been used; therefore, the information on performance for each device is limited. The current study is a comprehensive post‐market analysis of all the major functions and adverse events rates of the Xenios console and XLung kit 230 preconfigured ECMO tubing set in post‐cardiotomy applications.

Methods

This retrospective, single‐center analysis, conducted between 2017 and 2021, included 47 adult patients who received V‐A ECMO support using the Xenios system following cardiac surgery at our institution.

Results

The Xenios device was found to effectively improve and maintain microcirculation, organ perfusion, gas exchanges, organ function support, and stabilization of the cardiocirculatory system, while maintaining pump performance and stable body core temperature. Device‐related adverse events occurred in 28% of patients but resolved in 81% of cases without leaving clinical sequelae. Fewer hemolysis events than reported in the literature were recorded. Every patient experienced at least two patient‐related adverse events, but 43% had no relation to the device and 45% resolved without permanent damage.

Conclusions

The performance of the evaluated device was demonstrated in all key clinical areas, with an adequate safety profile.

This study is a comprehensive post‐market analysis of the Xenios console and XLung kit 230 preconfigured ECMO tubing set in post‐cardiotomy applications. The performance of this device was evaluated and demonstrated in all key clinical areas, with an adequate safety profile.

Abbreviations

AE

adverse event

AI

aortic valve insufficiency

AKI

acute kidney injury

ALAT

alanine aminotransferase

ALF

acute liver failure

ASAT

aspartate aminotransferase

BGA

blood gas analysis

Bpm

beats per minute

CABG

coronary artery bypass grafting

CPB

cardiopulmonary bypass

CPR

cardiopulmonary resuscitation

CRP

C reactive protein

CRT

cardiac resynchronization therapy

CT

computer tomography

DeltaP

ECMO transmembrane pressure

DP

arterial diastolic pressure

DRAE

device‐related adverse event

ECLS

extracorporeal life support

ECMO

extracorporeal membrane oxygenation

ELSO

Extracorporeal Life Support organization

fHB

free hemoglobin

FiO₂

fraction of inspired O₂

GI

gastrointestinal bleeding

HB

hemoglobin

HR

heart rate

ICB

intracranial bleeding

ICD

implantable cardioverter defibrillator

ICU

intensive care unit

IU

international units

LDH

lactate dehydrogenase

MAP

mean arterial pressure

MI

mitral valve insufficiency

MOF

multi organ failure

MRI

magnetic resonance imaging

MUV

Medical University of Vienna

MV

mechanical ventilation

O₂

oxygen

OHT

orthotopic heart transplantation

P1

ECMO drainage pressure

P2

ECMO pump pressure

P3

ECMO reperfusion pressure

PC

post‐cardiotomy

pCO₂

(arterial) carbon dioxide partial pressure

PCS

post‐cardiotomy syndrome

PEEP

positive end‐expiratory pressure

PGD

primary graft dysfunction

PM

pacemaker

P _mean

mean airway pressure

PNP

peripheral neuropathy

PNX

pneumothorax

pO₂

(arterial) oxygen partial pressure

PRAE

patient related adverse event

Rpm

rotations per minute

SAE

serious adverse event

SaO₂

arterial oxygen saturation

SGD

secondary graft dysfunction

SP

arterial systolic pressure

V‐A ECMO

veno‐arterial extracorporeal membrane oxygenation

ΔP

ECMO transmembrane pressure

1. Background

The Xenios console with the XLung kit 230 (Xenios AG, Germany) is widely utilized globally and represents the most commonly employed ECMO system in our institution (Department of Cardiac and Thoracic Aortic Surgery, Medical University of Vienna) at the time of writing. During the last decades, the use of ECMO treatment has constantly been increasing [1, 2] with a particular rise during the COVID‐19 pandemic [3]. Among the most common applications, patients requiring ECMO support after cardiac surgery (post‐cardiotomy ECMO [PC‐ECMO]) are clinically one of the most complex subgroups of patients. Because of their high morbidity and mortality, these patients require additional and special care [4, 5, 6]. Their situation is often critical due to the underlying heart disease, the high prevalence of comorbidities [7], the reduced cardiac function, the frequently unstable hemodynamic conditions, and the recent surgery [8] with concomitant use of cardiopulmonary bypass, often including cardioplegic arrest of the heart.

The Xenios XLung kit 230 is a preconfigured tubing set, which can be used with the Xenios console. With the appropriate tubing kit, the console is also capable of providing pulsatile blood flow [9]. The XLung kit consists of 3/8′′ blood tubing integrated with a Deltastream DP3 pump head (featuring mechanical bearings and magnetic coupling), built‐in pressure sensors, and a preconnected membrane lung (oxygenator). All components in contact with the patients' blood are coated with the proprietary x.ellence coating, a multilayer coating of heparin and albumin, except for parts of the DP3 pump. The set is approved for blood flows between 1 and 7 L/min and for continuous use up to 29 days.

Due to the increasing use of V‐A ECMO, the number of available devices is rising, yet the information on the safety and efficacy profiles of each console and tubing set is limited [9]. This study will evaluate clinical key domains such as end‐organ perfusion, hemodynamic function and gas exchange, besides the occurrence of complications to evaluate the clinical trajectory of patients after cardiac surgery and treated with the Xenios system.

2. Materials and Methods

2.1. Study Population

All 47 consecutive patients that were treated with the Xenios console and Xlung kit 230 after cardiac surgery between November 2017 and May 2021 at the Department of Cardiac and Thoracic Aortic Surgery of the Medical University of Vienna were included in this retrospective, single‐center post‐market analysis. The study protocol of our post‐cardiotomy ECMO database and its subsequent analysis is in line with the Declaration of Helsinki and was approved by our Institutional Review Board (EK No. 1086/2019, with a waiver for informed consent). Inclusion and exclusion criteria for our post‐cardiotomy database have been previously described in detail [10].

2.2. Endpoints

To investigate the safety of this device, the patient's hospital charts were thoroughly reviewed for any possible device or patient related adverse events (AE) from the start of ECMO support to hospital discharge. Bleeding events were assessed according to the registry definitions of the Extracorporeal Life Support organization (ELSO) [11]. In‐hospital mortality and hospital‐length‐of‐stay were collected for all patients.

The performance of the device was evaluated for:

• Improvement and maintenance of microcirculation and organ perfusion (decrease in lactate).

• Improvement and maintenance of gas exchanges (BGA).

• Stabilization of the cardiocirculatory system (improvement of vital signs and decreased need for vasoactive medication).

• Maintenance of ECMO performance (ECMO pump parameters).

• Maintenance of body core temperature (temperature in °C).

• Enabling and maintenance of lung‐protective ventilation (ventilator settings).

• Freedom from hemolysis (fHB < 50 mg/dL).

• Maintenance of balanced coagulation state (D‐Dimer and platelet count).

• Support of organ function (creatinine, ALAT, ASAT).

• Generating low systemic inflammation (CRP, leukocyte count).

Laboratory parameters, vital signs, respiratory parameters, and vasoactive medications were appraised immediately before ECMO implantation and once daily for every day on support with the Xenios device. The baseline of pump parameters was assessed 2–4 h after starting ECMO support. Lactate clearance was defined as $\frac{{\text{Lactate}}_{\mathrm{Pre}\text{implant}}-{\text{Lactate}}_{\mathrm{Day}1}}{{\text{Lactate}}_{\mathrm{Pre}\text{implant}}}*100\%$ [12]. Due to a change in our laboratory protocols, fHB was not available after November 2018, therefore the analysis of fHB was performed in only 19 patients that had values available.

To maintain data comparability, only the first ECMO run and the first cannulation strategy were included in the analysis.

2.3. Statistical Methods

The Kaplan–Meier method was used to estimate survival probabilities. For every endpoint the value assessed on the last day on the device was compared with the values available immediately before or after ECMO implantation, where appropriate. The two values were compared with the paired Wilkoxon signed rank test or paired t‐test if normal distribution was proven through Kolmogorov–Smirnov test. Results are shown as integers with percentage for categorical variables and median with interquartile range (Q1–Q3) or mean value with standard deviation if normal distribution was proven. Other specific analyses are described further in the text. Two‐sided p‐values < 0.05 were defined as statistically significant. Statistical analysis was performed with SPSS (IBM SPSS Statistics for Windows, Version 29.0.2.0 released 2023, Armonk, NY, USA) and R (R Core Team, version 4.3.2, released 2023).

3. Results

The study population consisted of 47 patients requiring postoperative ECMO support after cardiac surgery. Patient characteristics can be found in Table 1. The study collective was composed of particularly high‐risk patients, which had a median EuroScore II [13] of 14.32 (3.5–26.64)%, a CPB time of 266 (215–332) min and a median cross‐clamp time of 141 (91–179) min. ECMO support was initiated in 38 (81%) cases intraoperatively, while for the rest it was started in the ICU. Indication for ECMO implantation can be found in Table 1. The median duration of support was 135 (64–208) h [5.9 (2.7–8.7) days]. Index operations can be found in Table S1.

TABLE 1.

Patient characteristics.

Characteristics	Value
Age (years)	66.0 (55.0–72.0)
Height (cm)	172.5 (165.0–180.0)
Weight (kg)	80.0 (69.0–90.0)
Males	32 (68.1%)
ECMO indication
CPB weaning failure	18 (38%)
Acute lung edema	12 (26%)
PGD/SGD	8 (17%)
RV failure	3 (6%)
Arrhythmia	2 (4%)
Septic shock	2 (4%)
Post‐cardiotomy syndrome	2 (4%)
ECMO timing
Intraoperative	38 (81%)
Postoperative on ICU	9 (19%)
EuroScoreII	14.32 (3.5–26.64)
CPB duration (min)	266 (215–332)
Cross‐clamp time (min)	141 (91–179)
Cold ischemic time (OHT) (min)	207 (158–268)
Time on MV prior to ECMO implant (days)	0.33 (0.27–0.53)
Total time on MV (days)	10.38 (6.40–23.00)
Duration of ECLS (days)	5.87 (2.69–8.67)
Duration of ICU stay (days)	18.0 (10.0–39.0)
Duration of hospital stay (days)	34.5 (15.0–63.3)
In‐hospital mortality	22 (46.8%)
Multiorgan failure	12 (54.5%)
Intracranial bleeding	3 (13.6%)
Persisting heart failure	3 (13.6%)
Bleeding	1 (4.5%)
Ischemic stroke	1 (4.5%)
Cardiogenic shock	1 (4.5%)
Septic shock	1 (4.5%)

Abbreviations: CPB, cardiopulmonary bypass; ECLS, extracorporeal life support; ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit; MV, mechanical ventilation; OHT, orthotopic heart transplantation; PGD, primary graft dysfunction; RV, right ventricle; SGD, secondary graft dysfunction.

Arterial cannulation was mainly (62%) achieved through the subclavian artery with an 8 mm Polyethylene Terephthalate prosthesis sewn end‐to side to the artery and a vessel loop on the distal side of the artery to control blood flow to the arm and avoid hyperfusion, which is the standard of care at our center for post‐cardiotomy patients. The remaining patients were cannulated through a standard femoral cannulation (34%), an ECMO cannula to the ascending aorta (2%), or through a side branch of an ascending aortic prosthesis (2%) (Table 2).

TABLE 2.

ECMO discontinuation causes and cannulation sites.

Variables	Value
ECMO discontinuation reason
Expected recovery	32 (68.1%)
ECMO mortality	9 (19.1%)
Poor prognosis	3 (6.4%)
ECMO adverse event	2 (4.3%)
Bleeding control	1 (2.1%)
Arterial cannulation site
A. Subclavia	29 (61.7%)
Right	24 (82.8%)
Left	5 (17.2%)
A. Femoralis	16 (34%)
Right	12 (75%)
Left	4 (25%)
Aorta ascendens	1 (2.1%)
Central through side branch of ascendens prosthesis	1 (2.1%)
Venous cannulation site
V. Femoralis	44 (93.6%)
Right	33 (75%)
Left	11 (25%)
Right atrium	2 (4.3%)
Superior vena cava	1 (2.1%)

Note: Only the first cannulation strategy has been considered.

Abbreviations: A, arteria; ECMO, extracorporeal membrane oxygenation; V, vena.

3.1. Outcome

The in‐hospital survival rate in this collective was 53%. The median hospital length‐of‐stay from ECMO implantation until discharge was 35 (15–63) days. The most frequent cause of death was multiorgan failure (54.5%), followed by intracranial bleeding (13.6%), persisting heart failure (13.6), bleeding (4.5%), stroke (4.5%), cardiogenic shock (4.5%) and septic shock (4.5%) (Table 1). Of the 22 deaths that occurred, 68% occurred during ECMO support and 32% after ECMO weaning, during hospital stay. The Kaplan‐Maier analysis can be found in Figure 1.

FIGURE 1.

Kaplan–Meier estimates for in‐hospital survival.

3.2. Improvement and Maintenance of Microcirculation and Organ Perfusion

Variations of lactate levels showed a significant reduction of 2.2 (1.1–4.8) mmol/L [p < 0.001; pre ECMO 4.6 (2.9–7.0) mmol/L; last day of treatment 1.4 (1.0–2.0) mmol/L] in lactate levels from immediately before ECMO implantation and the end of support (Table 3). Lactate levels during treatment dropped already after 1 day of treatment to 1.7 (1.2–2.5) mmol/L (p < 0.01). Lactate clearance had a median of 8.1% (−21.1% to 41.3%) and the median duration to achieve concentrations at a threshold of < 2.0 mmol/L was 2 (2–3) days, where 9 (19.1%) patients never had values below the threshold and 5 (10.6%) patients had values < 2.0 mmol/L pre implant which rose during treatment.

TABLE 3.

Changes of the main outcome parameters on ECMO treatment.

Parameters	Mean/median pre‐ECMO a	Mean/median last treatment day	Mean/median of the differences	p
Microcirculation and organ perfusion
Lactate (mmol/L)	4.6 (2.9–7.0)	1.4 (1.0–2.0)	−2.2 (−4.8 to −1.1)	< 0.001*
Gas exchanges
pO2 (mmHg)	143.1 (±55.0)	109.3 (±29.1)	−34.7 (±60.4)	< 0.001*
pCO2 (mmHg)	38.2 (34.8–42.7)	36.1 (32.4–39.8)	−4.4 (−7.9 to 4.4)	0.097
SaO2 (%)	98.9 (95.9–99.5)	98.3 (97.2–98.9)	−0.5 (−1.7 to 2.8)	0.893
O2 content (mmHg)	431.2 (±167.2)	304.1 (±88.5)	−127.1 (±136.8)	0.017*
pH (0.00)	7.318 (±0.086)	7.435 (±0.083)	0.117 (±0.015)	< 0.001*
Vital signs and vasoactive drugs
SP (mmHg)	75 (±20)	104 (±22)	30 (±28)	< 0.001*
DP (mmHg)	52 (±11)	63 (±9)	11 (±15)	< 0.001*
MAP (mmHg)	60 (±12)	74 (±10)	14 (±16)	< 0.001*
Heart rate (bpm)	96 (82–105)	89 (80–105)	−1 (−13 to 7)	0.504
Noradrenalin (μg/kg/min)	0.260 (0.111–0.417)	0.067 (0.014–0.208)	−0.175 (−0.299 to −0.023)	< 0.001*
Dobutamine (μg/kg/min)	3.86 (±2.58)	2.94 (±1.75)	−0.92 (±2.68)	0.023*
Vasopressin (IU/h)	1.98 (0–4.02)	0 (0–2)	−1.00 (−3.00 to 0.00)	< 0.001*
Levosimendan (μg/kg/min)	0 (0–0.1)	0 (0–0)	0.0 (−0.1 to 0.0)	< 0.001*
Pump parameters
Pump speed (rpm)	6800 (6075–7400)	6000 (5400–6900)	−500 (−1250 to 25)	< 0.001*
Blood flow (L/min)	3.00 (2.28–3.50)	2.32 (2.03–2.96)	−0.32 (−1.11 to −0.07)	0.003*
Drainage pressure (mmHg)	−37 (−69 to −18)	−26 (−44 to −5)	12 (−4 to 35)	0.099
Pump pressure (mmHg)	210 (±48)	204 (±46)	−5.6 (±52.5)	0.678
Reperfusion pressure (mmHg)	179 (±43)	169 (±53)	−11 (±48)	0.393
Transmembrane pressure‐ΔP (mmHg)	32 (21–40)	25 (19–30)	−4 (−15 to 10)	0.515
FiO2 (%)	80 (60–100)	50 (44–60)	−20 (−38 to 0)	< 0.001*
Gas flow (L/min)	2.0 (1.5–2.5)	1.5 (1.0–2.0)	−0.5 (−1.3 to 0)	0.004*

Abbreviations: DP, arterial diastolic pressure; FiO₂, fraction of inspired O₂; MAP, mean arterial pressure; O₂, oxygen; pCO₂, (arterial) carbon dioxide partial pressure; pO₂, (arterial) oxygen partial pressure; SaO₂, arterial oxygen saturation; SP, arterial systolic pressure.

^a Pre‐ECMO values are the last values before ECMO implantation for all values except for pump parameters, where it corresponds to 2–4 h after implant.

^*Statistically significant (p < 0.05).

3.3. Improvement and Maintenance of Gas Exchange

A slightly low pH [7.318 (±0.086)] observed prior ECMO implantation increased by 0.117 (±0.015; p < 0.001) during ECMO support, reaching physiological values [7.435 (±0.083)] by the final day of treatment. Supranormal pO₂ values before ECMO started were found [143.1 (±55.0)]. During ECMO treatment, arterial pO₂ decreased by 34.7 (±60.4; p < 0.001) mmHg, bringing it to a normal value [109.3 (±29.1) mmHg] on the patients' last treatment day (Table 3).

3.4. Stabilization of the Cardiocirculatory System

The ability of the Xenios XLung kit 230 in stabilizing the cardiocirculatory system was assessed by evaluating the changes in arterial pressure and the need for vasoactive drugs. Arterial pressures increased significantly on the last treatment day [SP 30 (±28) mmHg (p < 0.001); DP 11 (±15) mmHg (p < 0.001); MAP 14 (±16) mmHg (p < 0.001)] compared with the values before treatment start, while the heart rate did not change significantly. The need for vasoactive medication decreased significantly with the use of the Xenios XLung kit 230 [Noradrenalin −0.175 (−0.299 to −0.023) μg/kg/min (p < 0.001); Dobutamine −0.92 (±2.68) μg/kg/min (p = 0.023); Vasopressin −1.00 (−3.00 to 0.00) IU/h (p < 0.001)], supporting effectively the cardiocirculatory system (Table 3).

3.5. Maintenance of ECMO Performance

Pump pressures, pump speed, and blood flow of the last day were compared with the values registered after 2–4 h after ECMO start. Regardless of outcome, pump speed and blood flow could be reduced [pump speed −500 (−1250 to 25) rpm (p < 0.001); blood flow −0.32 (−1.11 to −0.07) L/min (p = 0.003)] (Table 3) showing that weaning is possible. Gas flow and FiO₂ could also be reduced significantly [gas flow −0.5 (−1.3 to 0) L/min (p = 0.004); FiO₂ −20 (−38 to 0)% (p < 0.001)] indicating a tendency for a recovery of respiratory functions of these patients (Table 3).

3.6. Maintenance of Body Core Temperature

Maintenance of body core temperature was assessed every treatment day by measuring the difference between the set temperature on the Heater‐Cooler unit and the actual body core temperature. In this study population, the median body core temperature was 36.7 (36.4–36.9)°C and the Xenios XLung kit 230 was able to maintain a median difference of −0.4 (−0.8 to 0)°C, taking into account 255 cumulative treatment days (88%) during which both values were available.

3.7. Enabling and Maintenance of Lung‐Protective Ventilation (Ventilator Settings)

During ECMO treatment, ventilator settings were adapted to slowly wean the patient from support. We observed a median increase in PEEP values of 3 (0–4) mbar (p < 0.001), while the tidal volume did not change. All other respiratory parameters did not show significant changes (Table 4).

TABLE 4.

Changes of the main outcome parameters on ECMO treatment—Part 2.

Parameters	Mean/median pre‐ECMO a	Mean/median last treatment day	Mean/median of the differences	p
Respirator settings
P mean (mbar)	12.1 (±5.2)	12.7 (±2.7)	0.6 (±4.9)	0.403
Minute ventilation (L/min)	6.15 (4.55–8.2)	5.70 (3.85–7.15)	0.005 (−2.4 to 1.5)	0.474
Tidal volume (mL)	481.38 (±208.41)	479.89 (±171.05)	−1.49 (±229.19)	0.966
Respiratory rate (bpm)	14 (10–16)	12 (10–15)	0 (−4 to 2)	0.457
PEEP (mbar)	5 (5–7)	8 (7–10)	3 (0 to 4)	< 0.001*
Coagulation state
D‐Dimer (μg/mL)	62.53 (±83.87)	9.12 (±7.26)	−53.41 (±91.13)	0.559
Platelets (G/L)	181 (±88)	60 (±34)	−121 (±88)	< 0.001*
Fibrinogen (mg/dL)	352 (±132)	338 (±149)	−14 (±197)	0.663
Antithrombin III Akt (%)	73.56 (±27.95)	62.93 (±23.8)	−10.63 (±35.28)	0.116
Organ function
Kidney function
Urea (mg/dL)	23.7 (14.4–40.4)	32.9 (22.7–50.7)	−0.13 (−0.51 to 0.49)	0.629
Creatinine (mg/dL)	1.22 (0.94–1.85)	1.28 (0.90–1.94)	7.2 (−7.0 to 20.3)	0.111
Liver function
ASAT (U/L)	33.0 (21.0–165.0)	73.5 (34.0–142.3)	9.7 (−90.0 to 44.5)	0.597
ALAT (U/L)	29.0 (19.8–95.5)	22.5 (12.0–71.5)	−7.0 (−25.5 to 5.5)	0.03*
Inflammation marker
CRP (mg/dL)	3.59 (±4.92)	14.68 (±9.97)	11.09 (±12.12)	< 0.001*
Leukocytes (G/L)	9.09 (7.29–15.30)	10.49 (7.19–14.93)	2.04 (−4.37 to 4.42)	0.408

Abbreviations: ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; CRP, C reactive protein; PEEP, positive end‐expiratory pressure; P _mean, mean airway pressure.

^a Re‐ECMO values are the last values before ECMO implantation for all values except for pump parameters, where it corresponds to 2–4 h after implant.

^*Statistically significant (p < 0.05).

3.8. Generating Low Hemolysis

The occurrence of hemolysis was evaluated by changes in free Hemoglobin (fHb), Hemoglobin, Hematocrit, LDH, and Bilirubin (Table 5). Free hemoglobin was considered the most significant indicator for hemolysis. Free hemoglobin had a mean value of 7.7 (±13.09) mg/dL. Out of 133 (40%) daily available values, only 3 (2%) indicated significant hemolysis with values over 50 mg/dL, and 16 (12%) revealed mild to moderate hemolysis with values comprised between 10 and 50 mg/dL. The vast majority of values [114 (86%)] were under the threshold of 10 mg/dL. All patients had at least one measurement under 10 mg/dL, while 8 (42%) patients had one or more episodes of mild/moderate hemolysis and 3 (16%) patients had one event with fHB over 50 mg/dL, with the highest reported value of 103.14 mg/dL.

TABLE 5.

Mean and maximal values of hemolysis parameters.

Parameters	n = 335	Mean (±SD)	Max. value
Free hemoglobin (mg/dL)	133 (39.7%)	7.7 (±13.09)	103.14
Hemoglobin (g/L)	335 (100%)	9.1 (±1.8)	28.3
Hematocrit (%)	334 (99.7%)	26.5 (±4.7)	46
LDH (U/L)	316 (94.3%)	774 (±771)	6630
Bilirubin (mg/dL)	326 (97.3%)	3.73 (±4.49)	19.65

Abbreviations: LDH, lactate dehydrogenase; SD, standard deviation.

3.9. Maintenance of Balanced Coagulation State

During ECMO support platelets decreased significantly by −121 (±88) G/l (p < 0.001), which may indicate some damaging or consumptive process by the pump or the tubing set (Table 4). Fibrinogen and D‐Dimer levels did not significantly change during the observation period.

3.10. Support of Organ Function

Kidney function was assessed by evaluating the changes in plasmatic concentration of creatinine and urea between the baseline and the last day of support. Kidney function did not change significantly. The median creatinine during support was 1.34 mg/dL (0.97–1.94 mg/dL), and therefore slightly elevated. However, 21 patients (45%) needed renal replacement therapy. A statistically significant decrease in ALAT concentrations was also evident [−7.0 (−25.5 to 5.5) U/L; p = 0.03] (Table 4).

3.11. Infection/Inflammatory State

It is known that extracorporeal circulation causes increased systemic inflammation and patients on support with the Xenios XLung kit 230 experienced an increase in CRP levels of 11.09 (±12.12) mg/dL. Increase in leukocyte count was not statistically significant (Table 4). There was no statistical difference in CRP levels between patients that had a clinically manifested infection and those who did not [infection: 11.13 (5.15–19.44) mg/dL; no infection: 13.46 (5.93–19.13) mg/dL; p = 0.511].

3.12. Device‐Related Adverse Events (DRAEs)

DRAEs were collected for all patients (Table 6). A total of 16 DRAEs were registered in 13 patients (28%), which resulted in a per‐patient event rate of 0.34. The most common DRAEs were technical malfunctions of device components [8 (17%)] followed by thrombotic events [4 (8.5%)]. In 3/8 (38%) of the system malfunctions, an excessive pump vibration or noise was reported, which could be resolved by reducing the pump speed in one case; in the remaining two cases, no measures were taken and no clinical consequence arose. The excessive noise and vibration could be explained by an imbalanced impeller, caused by fibrinous or thrombotic depositions. In two cases (4% of all patients), a software or sensor malfunction occurred, where data on blood flow and pressures were not shown anymore. In one case, the problem persisted after a flow sensor exchange and a system exchange was needed. This event had no clinical consequences for the patient. In the tubing set of two patients (4%), some thrombotic material was found. The solution to this DRAE was in one case a system exchange, without negative impact on the patient, while in the other case only a part of the circuit was exchanged. Two system changes were performed, but the reason for the exchange was not documented. Both underwent the procedure without adverse clinical consequences for the respective patient.

TABLE 6.

Device related adverse events (DRAE).

DRAE	Frequency (n = 16)	Patients affected (n = 47)
System malfunction
Pump head	1 (6.25%)	1 (2.1%)
Heat exchanger	1 (6.25%)	1 (2.1%)
No display of flow and pressure data (sensor/software malfunction)	2 (12.5%)	2 (4.3%)
Excessive pump vibration or noise	3 (18.75%)	3 (6.4%)
Sudden pump‐stop when patient coughed	1 (6.25%)	1 (2.1%)
Thrombotic complication
Thrombotic material in tubing set	2 (12.5%)	2 (4.3%)
High D‐Dimer	1 (6.25%)	1 (2.1%)
Oxygenator thrombosis	1 (6.25%)	1 (2.1%)
Other kinds of complication
Leakage of the subclavian anastomosis	1 (6.25%)	1 (2.1%)
Air in the venous line	1 (6.25%)	1 (2.1%)
Unknown	2 (12.5%)	2 (4.3%)
Resolution
Resolved without sequelae	13 (81.3%)	11 (23.4%)
Resolved with possible sequelae	1 (6.3%)	1 (2.1%)
Unknown	2 (12.5%)	2 (4.3%)

Abbreviation: DRAE, device related adverse events.

A total of six system changes were performed, accounting for 60% of all the actions taken to resolve the DRAE and were necessary in 12.8% of patients. System changes were performed in a median of 10.5 (1.0–13.25) days after ECMO start. The reasons for system exchanges were a system thrombosis (16% of exchanges), one oxygenator failure (16%), high D‐dimer (16%), one pump head malfunction (16%), and in one case air entered the venous line of a stopped ECMO during a surgical revision on CPB (16%). The remaining system exchange was due to an undocumented event (16%).

No death resulted from the described DRAEs, and in 81% of the cases, the problem resolved without leaving clinical sequelae. Only one pump head malfunction and subsequent CPR and system exchange could possibly be tied to an unclear neurologic deficit.

3.13. Patient‐Related Adverse Events (PRAEs)

Post‐cardiotomy patients that are in need of ECMO support are generally a particularly morbid collective of patients where adverse events are frequent and well described in the literature [14].

A total of 307 PRAEs have been collected, 43% of which had no relation to the device, 45% resolved without leaving consequences for the patient, and 22% resolved without any need for intervention. Every patient experienced at least 2 PRAEs (Table 7).

TABLE 7.

Patient related adverse events.

PRAE	n = 307
Pro‐patient event rates
Minimum	2
Median (IQR)	6 (4–8.5)
Maximum	14
Timing
On ECMO	168 (54.7%)
Post ECMO	136 (44.3%)
Unknown timing	3 (1%)
AE start (day), median (IQR)	4.5 (1–12)
AE stop (day), median (IQR)	10 (4–19)
AE duration (days), median (IQR)	0 (0–7)
Relation to the device
No correlation	132 (43%)
Not to be excluded	95 (30.9%)
Possible	34 (11.1%)
Probable	13 (4.2%)
Definite	32 (10.4%)
Unknown relation	1 (0.3%)
Action taken
No action taken	68 (22.1%)
Surgical revision/intervention	57 (18.6%)
Pharmacological therapy	107 (34.9%)
Renal replacement therapy	21 (6.8%)
Tracheostomy	7 (2.3%)
CPR	12 (3.9%)
PM/ICD/CRT implantation	3 (1%)
Physical therapy	4 (1.3%)
Cardioversion	10 (3.3%)
Other treatment	6 (2%)
Unknown	12 (3.9%)
Severity
AE	22 (7.2%)
SAE	285 (92.8%)
Resolution
Resolved without sequelae	139 (45.3%)
Resolved with sequelae	45 (14.7%)
Unresolved	23 (7.5%)
Unknown	26 (8.5%)
Death	74 (24.1%)

Abbreviations: AE, adverse event; CRP, C reactive protein; CRT, cardiac resynchronization therapy; ECMO, extracorporeal membrane oxygenation; ICD, implantable cardioverter defibrillator; PM, pacemaker; PRAE, patient related adverse event; SAE, serious adverse event.

Neurological PRAEs occurred frequently during and after ECMO support. 57% of patients experienced at least one neurological PRAE, but only 23% of patients experienced an acute brain injury (defined as any CT/MRI proven ischemic stroke, intracranial/subarachnoid/subdural hemorrhage of any severity) (Table 8). Of the 11 reported acute brain injuries, 27% resolved with remaining sequelae, while 64% resulted in or contributed to the patient's death.

TABLE 8.

Main patient related adverse events (PRAE).

PRAE categories	n (%)	Patients affected (n = 47)
Neurologic PRAE	35 (100%)	27 (57.4%)
Acute brain injury	11 (31.4%)	11 (23.4%)
Postoperative delirium	14 (40%)	14 (29.8%)
Polyneuropathy	6 (17.1%)	6 (12.8%)
Seizures	4 (11.4%)	3 (6.4%)
Hemorrhagic PRAE	30 (100%)	24 (51.1%)
Surgical site bleeding	10 (33.3%)	10 (21.3%)
Cannulation site bleeding	8 (26.7%)	8 (17%)
Hemothorax	7 (23.3%)	7 (14.9%)
GI bleeding	1 (3.3%)	1 (2.1%)
Retroperitoneal hematoma	1 (3.3%)	1 (2.1%)
Bleedings in other locations	3 (10%)	3 (6.4%)
Limb PRAE	13 (100%)	12 (25.5%)
Compartment syndrome	4 (30.8%)	4 (8.5%)
Seroma	3 (23.1%)	3 (6.4%)
Thrombosis	3 (23.1%)	3 (6.4%)
Ischemia	2 (15.4%)	2 (4.3%)
Pseudoaneurysm	1 (7.7%)	1 (2.1%)
Infection PRAE	50 (100%)	29 (61.7%)
Pneumonia	16 (32%)	16 (34%)
Cannulation site/surgical wound infection	7 (14%)	7 (14.9%)
Sepsis/septic shock	6 (12%)	6 (12.8%)
Positive blood cultures	9 (18%)	9 (19.1%)
Catheter associated infection	4 (8%)	3 (6.4%)
Sternal infection	2 (4%)	2 (4.3%)
Urinary tract infection	2 (4%)	2 (4.3%)
Ruptured septic pseudoaneurysm at cannulation site	1 (2%)	1 (2.1%)
Infected pacemaker lead	1 (2%)	1 (2.1%)
Other infection	2 (4%)	2 (4.3%)
Respiratory PRAE	14 (100%)	11 (23.4%)
Pneumothorax	3 (21.4%)	3 (6.4%)
Prolonged respiratory weaning	5 (35.7%)	5 (10.6%)
Respiratory failure	2 (14.3%)	2 (4.3%)
Lung bleeding	2 (14.3%)	2 (4.3%)
Bronchospasm	1 (7.1%)	1 (2.1%)
Other	1 (7.1%)	1 (2.1%)
Cardiovascular PRAE	8 (100%)	8 (17%)
Acute valve insufficiency	2 (25%)	2 (4.3%)
Right heart failure	1 (12.5%)	1 (2.1%)
Pericardial tamponade	1 (12.5%)	1 (2.1%)
Myocardial infarction	2 (25%)	2 (4.3%)
Cardiac decompensation	1 (12.5%)	1 (2.1%)
Left ventricle distention	1 (12.5%)	1 (2.1%)

Abbreviations: GI, gastrointestinal bleeding; PRAE, patient related adverse event.

Bleeding events accounted for 10% of all PRAEs and were experienced by 51% of patients (Table 8). Of these 30 events, 10 (21% of patients) were surgical site bleedings, 8 cannulation bleedings (17% of patients), seven hemothorax (15% of patients), one gastrointestinal bleeding (2% of patients), one retroperitoneal hematoma (2% of patients), and three bleedings in other locations (2% of patients). Only one surgical site bleeding in this collective led or contributed to the patient's death.

A total of 13 limb adverse events were documented in 12 patients (26%) (Table 8). Of these 13 PRAE, four (9%) were compartment syndromes, three (6%) seromas, three (6%) arterial thrombosis, two (4%) limb ischemia, and one (2%) post removal pseudoaneurysm.

During and after ECMO support, infections were a frequent event. 62% of the patients were affected at least once by an infection (Table 8). Pneumonia was the most frequent infection (34%), followed by positive blood cultures (19%) and manifest sepsis or septic shock (13%). Infections of the cannulation site or the surgical wound occurred in 15% of patients, and two patients (4%) developed a sternal infection. Additionally, one ruptured septic pseudoaneurysm of the cannulation site was reported.

In total, 11% of the patients had prolonged weaning from the ventilator, 6% developed a pneumothorax, 4% had respiratory failure, and 4% had lung bleeding (Table 8).

Cardiovascular PRAEs occurred in 19% of these patients (Table 8). The most frequent were acute valve insufficiencies (4%), myocardial infarctions (4%), and pericardial tamponades (4%). Other PRAEs are described in Table 9.

TABLE 9.

Patient related adverse events (PRAE).

PRAE	n = 307	Patients affected (n = 47)
Death	22 (7.2%)	22 (46.8%)
Neurologic (ICB, stroke, seizures, PNP, delirium, etc.)	35 (11.4%)	27 (57.4%)
Bleeding (cannulation site, hemothorax, tamponade, GI, etc.)	30 (9.8%)	24 (51.1%)
Cardiovascular (acute decompensation, acute AI, acute MI, acute right heart failure, tamponade, CABG grafts closure, etc.)	8 (2.6%)	8 (17%)
CPR	11 (3.6%)	10 (21.3%)
Arrhythmias	28 (9.1%)	24 (51.1%)
Respiratory (prolonged respiratory weaning, PNX, etc.)	14 (4.6%)	11 (23.4%)
MOF	8 (2.6%)	8 (17%)
Infections (pneumonia, bloodstream i., sepsis, etc.)	50 (16.3%)	29 (61.7%)
Bowel ischemia/perforation	4 (1.3%)	4 (8.5%)
Metabolic (AKI, ALF, hemolysis, etc.)	65 (21.2%)	39 (83%)
Limb (compartment syndrome, acute/chronic vessel occlusion, ischemia, necrosis, seroma, etc.)	13 (4.2%)	12 (25.5%)
Harlequin syndrome (differential hypoxia)	2 (0.7%)	2 (4.3%)
Heparin induced thrombocytopenia	3 (1%)	3 (6.4%)
Graft Rejection (St. p. OHT)	2 (0.7%)	2 (4.3%)
Other AE (skin rush, lupus anticoagulants, rhabdomyolysis, spleen infarction, etc.)	12 (3.9%)	11 (23.4%)

Abbreviations: AI, aortic valve insufficiency; AKI, acute kidney injury; ALF, acute liver failure; CABG, coronary artery bypass grafting; CPR, Cardiopulmonary resuscitation; GI, gastrointestinal bleeding; ICB, intracranial bleeding; MI, mitral valve insufficiency; OHT, orthotopic heart transplantation; PNP, peripheral neuropathy; PNX, pneumothorax; PRAE, patient related adverse event.

4. Discussion

ECMO after cardiac surgery can be considered a last‐resort treatment and therefore patients undergoing this treatment are always highly morbid and unstable. A high in‐hospital mortality is expected, and the mortality reported in this study for patients treated with the Xenios XLung kit 230 is in line with the reports of similar patient collectives [10, 15]. Despite the markedly elevated EuroSCORE II of 14.32% (3.5–26.64), post‐weaning mortality was only slightly higher than that reported in the available literature for cohorts with comparable risk profiles (32% vs. 23.1%–30.7%) [15, 16].

The performance of this device was analyzed in‐depth through all the major aspects of ECMO support. The main goal of ECMO support is to improve and maintain organ perfusion and microcirculation. Despite the highly morbid patient collective the device proved to effectively improve these functions, and lactate levels decreased in the majority of patients to physiological values within 2 days of treatment.

The other main application of V‐A ECMO in post‐cardiotomy setting is the stabilization of the cardiocirculatory system. In this collective we could prove that throughout ECMO support, patient's arterial pressure values could be heightened even with a significant reduction of vasoactive medication dosage. This could be evidenced even in the 32% of patients that could not be weaned and died while on ECMO support. This indicates the effectiveness of the device in stabilizing the cardiocirculatory system, which allows for recovery and weaning from support in most cases.

Initially deranged values like pH and pO₂ can be explained by the fact that only post‐cardiotomy patients were included in this study, where often the baseline blood gas analyses were performed while the patients were still on cardiopulmonary bypass support, but returned to physiological levels on ECMO treatment proving that the device can effectively support, maintain, and improve gas exchange.

Assessment of the heat‐exchanger performance showed that the device could bring the body core temperature to a set value with a deviation of only −0.4 (−0.8 to 0)°C.

With the advent of magnetically driven pumps that rely on hydrodynamic bearings, hemolysis has become a less relevant factor, during the normal operation of the pump. In contrast, hemolysis can become an important marker for other problems, such as thrombotic events inside the oxygenator or tubing [17, 18]. Further causes for hemolysis in these patient collectives can be inadequate cannula positions, kinking or stenosis of grafts, surgical anastomosis or tubing. As far as can be evaluated with the limited data available on fHB, in our cohort only 2% of patients treated with the Xenios device experienced hemolysis events with fHB higher than 50 mg/dL, compared to the 3.4% reported by ELSO [5]. Our data suggests that the high rotational speed (up to 10 000 rpm) combined with the small priming volume inside the pump does not lead to increased blood damage in real life conditions. Computational fluid dynamics and in vitro studies of this pump have shown that at low flow conditions hemolysis increases sensibly [19], which could be relevant during the weaning phase and for pediatric applications, but these aspects could not be assessed due to the limited data availability of these parameters and need further investigation.

In this study, an impairment of the coagulation could be evidenced by a significant and progressive reduction of platelet count. This phenomenon could describe a destructive or consumptive process by the oxygenator membranes, the tubing surfaces, or the pump. This process has also been described previously in literature [20, 21]. It is to be noted that 3 (6.4%) of patients developed heparin‐induced thrombocytopenia and 24 (51.1%) experienced bleeding events that could have contributed to the drop in platelet count.

With regard to organ function, ECMO support resulted in an improvement of liver function shown by a significant decrease in ALAT concentrations. Kidney function was assessed by the difference in serum creatinine between the value that was taken on the last ECMO treatment day and the last value that was taken before ECMO start, which showed no significant difference. This analysis method does not discriminate against changes during support, nor does it take into account the fact that 45% of patients needed some form of renal replacement therapy.

A statistically significant rise in CRP levels was observed among our study collective, even in those that did not have a clinically evident infection. This rise in CRP could in part be caused by the interaction of the patients' blood with the synthetic surface of the device. This phenomenon is a well‐known adverse event and comprehensively documented in scientific literature [22].

DRAEs are often dramatic events that can endanger a patient's life. The high number of these events [16 (28% of all patients)] in this collective is easily explained by the fact that even DRAES with no or a low impact on the patient, such as high pump vibration or noise, were collected. High pump vibration or noise events could not be found in literature, possibly because of the benign outcome of this event or due to a specific faulty device unit at our center. Circuit clotting seems to be less frequent in our collective (12.5% of all DRAEs) than reported from ELSO registry [11] (18.4% of all HRAE).

The patient specific risk factors that favor the need for ECMO support after cardiac surgery are mainly the same that favor a complex postoperative convalescence with several adverse events. For this reason and the very low survival chance without ECMO support in these patients, a high number of adverse events is tolerated.

Every patient experienced at least two PRAEs, but 43% had no relation to the device and 45% resolved without permanent damage to the patient, 22% even without intervention.

In comparison with the PRAE rates found in the literature, our collective treated with a Xenios device had a lower acute brain injury rate (23%) than the 38% incidence reported by Wilcox et al. [23].

Compared to the results from a previous study from our group [24] bleedings in this collective seem to happen at a higher rate than in the general PC‐ECMO population (51% vs. 39%). Especially considering surgical site bleedings and hemothorax combined, patients treated with the Xenios device had more events [34% (16 patients) vs. 27%] and comparable cannulation site bleedings (17% vs. 13%). This result is unlikely to be associated with the used ECMO device; rather, plausible explanations are the sensibly smaller cohort or a particularly morbid collective.

This other study [24] reported a lower incidence of fasciotomies (5% vs. 9%) and a frequency of 10% of wound healing disorder and seromas combined. In our cohort, we registered only 6% of seromas, without considering wound healing disorders. Furthermore, our collective exhibited a reduced prevalence of limb ischemia (4%) in comparison to the 10% documented in the study. It is to be noted that these adverse events greatly depend on the cannulation strategy used, which could contribute to these differences.

Other publications report an infection rate of 8%–64% for all ECMO patients [25, 26, 27] with especially high rates in PC‐ECMO patients. Our study reports an incidence of 62%, which lies within the reported range. Compared to others [27] our study cohort had a lower incidence of pneumonia (34% vs. 55%) and slightly higher cannulation site infections (15% vs. 10%).

We performed a thorough analysis of the PRAEs and compared it to the scientific literature. Results showed that patients treated with the Xenios device had tendentially less acute brain injuries, pneumonia, seromas, or limb ischemia, but more bleeding events and compartment syndromes compared to other ECMO studies [23, 24, 27]. These differences were not statistically validated and tended to be small. Overall, this device seemed to show similar results when compared to other systems reported in the current literature.

5. Limitations

This study is limited by the retrospective, single‐center design. Although this ensured homogeneous treatment and data collection within the study population, it also reduces its generalizability. A relatively small sample size and limited data availability pose further limits.

Every variable was collected once daily, all at the same time, which gives a snapshot of a single moment of the day, but could, for some variables like blood pressure, not be representative for the rest of the day.

At our center in November 2018, our laboratory changed measurements of hemolysis from plasma free hemoglobin to hemolysis index, which correlates strongly with the previous measurements, but it is not standardized and therefore not comparable with values reported by other laboratories. We decided to assess only free hemoglobin, even if we had measurements for only 19 of the 47 patients, limiting the power of the analysis of this aspect.

The collection and analysis of the patient‐related adverse events was performed meticulously, but the confrontation with the current literature was made difficult by the lack in literature of comprehensive analysis of PRAEs, by the different definitions, data collection criteria, and different patient collectives. Some of the reported studies were performed at our center, with potential cross‐over of patients. Randomized trials are therefore necessary to effectively compare different devices.

6. Conclusions

We performed a comprehensive evaluation of all the fundamental aspects of a single ECMO device model, which showed that the Xenios console with the XLung kit 230 has a good performance and adverse event rates comparable to current literature. This device was found to effectively stabilize patients in all the major organ functions while maintaining ECLS performance and generating low hemolysis.

Author Contributions

Luca Conci: data collection, data analysis/interpretation, statistics, drafting article. Anne‐Kristin Schaefer: concept/design, critical revision of article, approval of article. Thomas Schlöglhofer: data analysis/interpretation, critical revision of article, approval of article. Emilio Osorio‐Jaramillo: critical revision of article, approval of article. Amila Kahrovic: approval of article. Alvaro Perazzo: critical revision of article, approval of article. Marcus Granegger: critical revision of article, approval of article. Daniel Zimpfer: concept/design, approval of article. Dominik Wiedemann: concept/design, funding secured by, critical revision of article, approval of article. Julia Riebandt: concept/design, critical revision of article, approval of article.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Operations performed before ECMO implantation.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.