Secondary Logo

Journal Logo

Clinical Critical Care

In Vivo Testing of Extracorporeal Membrane Ventilators: iLA-Activve Versus Prototype I-Lung

Kischkel, Sabine*†; Bergt, Stefan; Brock, Beate; von Grönheim, Johan*; Herbst, Anne*; Epping, Marc-Jonas*; Matheis, Georg§; Novosel, Esther§; Schneider, Joerg§; Warnke, Philipp; Podbielski, Andreas; Roesner, Jan P.‡‖; Lelkes, Peter I.#; Vollmar, Brigitte*

Author Information
doi: 10.1097/MAT.0000000000000465


Chronic obstructive pulmonary disease (COPD) is a major global health burden. Presently, COPD is the fourth leading cause of death worldwide, and is expected to progress to the third leading cause of death by 2030.1 Patients with severe COPD often suffer from acute respiratory failure owing to bacterial or viral infection of the lungs. Respiratory failure leads to life-threatening retention of carbon dioxide (CO2) and lack of oxygen (O2) in blood and tissues. Conventional clinical therapies include mechanical ventilation under anesthesia and administration of antibiotics in the intensive care unit (ICU). Extracorporeal membrane ventilators can achieve sufficient CO2 elimination in several clinical conditions of acute lung failure.2 A major advantage of using extracorporeal membrane ventilators is the option to avoid artificial mechanical ventilation and the associated necessary anesthesia. Patients are awake and can actively participate in a variety of therapeutic approaches, such as mobilization and physiotherapy. Because of their dimensions, systems such as iLA-activve allow some form of restricted mobilization; however, that mobility is limited to the ICU. Availability of a miniaturized and wearable extracorporeal ventilator could significantly improve the mobility of awake patients undergoing therapy for acute lung failure (bridge to recovery).

The primary aim of this study was to provide a first side-by-side comparison of the decarboxylation efficacy of a new miniaturized and transportable extracorporeal membrane ventilator prototype, I-lung, versus the commercial system iLA-activve for more than a period of 72 hours in a large animal model.

Materials and Methods

Animal experiments were approved by the governmental ethical board for animal research (Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei, Mecklenburg-Vorpommern, Germany; No: 7221.3-1.1-021/14) and were carried out in accordance with the EU-directive 2010/63/EU. Fifteen young adult German Landrace pigs with a mean body weight (BW) of 50.1 ± 3.5 kg were the subjects of this study. The animals were housed at the Institute for Experimental Surgery (Rostock University Medical Center) and were given free access to standard laboratory chow and water.

Anesthesia and Instrumentation

After overnight fasting and receiving water ad libitum, pigs were premedicated intramuscularly with 5 mg/kg azaperone (Janssen-Cilag GmbH, Neuss, Germany), 45 mg/kg ketamine (Belapharm, Vechta, Germany), 0.2 mg/kg midazolam (Dormicum, Hoffmann La Roche AG, Grenzach-Wyhlen, Germany), and 0.01 mg/kg atropine (B. Braun, Melsungen AG, Melsungen, Germany). Animals were placed in supine position on a heating pad to keep their body temperature constant at 37°C. While monitoring O2 saturation and heart rate (HR), animals were preoxygenated with 100% O2. Anesthesia was induced intravenously via ear vein cannulation (18 G, B. Braun). After a bolus injection of 3 μg/kg fentanyl (Fentanyl-Janssen, Hanssen-Cilag, Neuss, Germany) and 2 mg/kg propofol (Propofol-Lipuro, B. Braun), as well as 0.4 mg/kg cisatracurium (Nimbex, GlaxoSmithKline GmbH & Co. KG, Munich, Germany), the pigs were intubated by tracheotomy (ID 8 mm Shiley Hi-Lo Tube, Mansfield, MA). Anesthesia was maintained by continuous intravenous infusion of 1.2–2.0 mg/kg/h propofol, 0.06–0.10 mg/kg/h midazolam, 1.5–2.5 mg/kg/h ketanestS, 0.6–1.8 μg/kg/h clonidine (Paracefan, Boehringer Ingelheim Pharma GmbH & Co. KG, Ingelheim, Germany), and 30–50 μg/kg/h fentanyl. A Servo300 ventilator (Siemens, Erlangen, Germany) was used to provide pressure-controlled mechanical ventilation. Respiratory rate and tidal volume were adjusted at the beginning of each experiment to maintain arterial partial pressure of CO2 (PaCO2) between 4 and 6 kPa. Inspiratory fraction of O2 (FiO2) was reduced from 1.0 to 0.4 and a positive end-expiratory pressure (PEEP) of 5 cm H2O was applied to maintain arterial partial pressure of O2 (PaO2) between 12 and 15 kPa during the preparation period of up to 2 hours.

A quad-lumen central venous catheter (8.5 Fr Arrow-Howes, CVC, Arrow International Inc., Reading, PA) was inserted via the left internal jugular vein and a 22 Fr double lumen catheter (NovaPort twin catheter, Novalung GmbH, Heilbronn, Germany) via the right jugular vein. The right femoral artery was cannulated with an arterial PiCCO-catheter (5 Fr thermodilution catheter, 20 cm, Pulsion Medical System, Feldkirchen, Germany) for continuous monitoring of hemodynamic parameters. Body temperature was monitored continuously with the PiCCO-catheter and kept at 37.0 ± 0.5°C. A urinary balloon catheter (Rüsch, Teleflex Medical, Kernen, Germany) was placed via mini-laparotomy into the urinary bladder to monitor urine output. For maintenance of normovolemia, all animals received an intravenous electrolyte solution (Sterofundin ISO 1/1E, B. Braun) at 6–10 ml/kg/h throughout the experiments. Additional fluid was infused, as necessary, according to the PiCCO-catheter hemodynamic measurements, with the following target values: mean arterial pressure (MAP) > 70 mm Hg, stroke volume variation (SVV) < 20%, extravascular lung water index (ELWI) > 10 ml/kg, and intrathoracic blood volume index (ITBVI) during 800–950 ml/m2.

After completion of the instrumentation, both extracorporeal membrane ventilators were filled with heparinized (1.000 IU) saline solution, connected to the 22 Fr double lumen catheters, and operated at an extracorporeal blood flow of 1.2 L/min and a sweep gas flow of 8 L/min. Throughout the study (up to 72 hours), the pressure gradient across the membrane was between 20 and 25 mm Hg, and the average drainage pressure was approximately −30 mm Hg.

Concomitantly with starting the two devices, animals were anticoagulated with 5.000 IU of heparin (Heparin-Sodium, B. Braun). This was followed by a continuous intravenous heparin infusion according to the activated clotting time (ACT), with the aim to maintain an ACT in the range of 150–250 seconds.

Extracorporeal Lung Assist Systems/Device Technology

For extracorporeal CO2 removal (ECCO2R) during hypoventilation-induced hypercapnia, functional prototypes of the newly developed miniaturized, wearable I-lung system (Figure 1) and the commercially available iLA-activve systems were used, side by side. Both devices are manufactured by Novalung (Xenios AG, Heilbronn, Germany). iLA-activve is an established extracorporeal lung assist device for the decarboxylation and oxygenation of blood and is routinely used in ICUs for extracorporeal membrane oxygenation (ECMO) and ECCO2R.3 In this study, iLA-activve was used as a benchmark4,5 and comparative system to the newly developed I-lung prototype. The core of the I-lung system is a miniaturized gas exchanger, which was designed for optimal blood flow and gas exchange characteristics while concomitantly reducing the size of the device. In both systems, the gas exchange occurs across poly-methylpentene (PMP) hollow fiber membranes. The blood-contacting surfaces were coated identically in both devices with a heparin-containing multilayer coating. In both systems, the two controllable variables for CO2 elimination, namely blood flow rate and sweep gas flow rate, were adjustable and maintained at identical levels. In addition, both systems are equipped with sensors for monitoring blood pressure, air-bubbles, blood flow rate, and sweep gas flow rate. Because I-lung uses ambient air as sweep gas, iLA-activve was also operated with ambient air, instead of using O2. The I-lung casing tested in this study is the first prototype of a single-use, disposable carrying device.

Figure 1.:
New I-lung mobile, wearable lung assist device. The dimensions of the case are approximately 220 × 310 × 105 mm3. The compact case contains blood tubes, particle filter, gas flowmeter, blood flow and bubble sensor, blood pump head, I-lung gas exchanger, sweep gas pump, silencer, and the I-lung blood pump drive.

Data Assessment

Heart rate, MAP, central venous pressure (CVP), cardiac output (CO), and SVV were assessed via intravascular catheters; CO was determined by thermodilution (mean of three injections of 15 ml ice-cold saline using the PiCCO-System). Arterial blood samples were taken every 3 hours for analyzing hemoglobin, hematocrit, arterial blood gases, acid base status, electrolytes, as well as glucose and lactate concentrations using an autoanalyzer (Radiometer, ABL 800, Copenhagen, Denmark). In addition, these blood samples were also analyzed for ACT (Medtronic Hemotec, Inc., Parker, Colorado). Arterial blood samples were analyzed every 24 hours for differential blood cell count (automated clinical analyzer KX-21, Sysmex Deutschland GmbH, Norderstedt, Germany) and for assessing transaminases (alanine transaminase [ALT], aspartate transaminase [AST]), hemolysis parameters (haptoglobin, free hemoglobin, lactate dehydrogenase [LDH]), and creatinine kinase. The latter analyses were performed by the central laboratory of Rostock University Medical Center. In addition, arterial blood samples were also analyzed for TNF-α, IL-1β, IL-6 (R&D Systems, Inc., Minneapolis, MN) and soluble P-selectin (Abbexa Ltd., Cambridge, UK) using the respective enzyme immunoassay kits according to the manufacturers’ instructions.

At the end of the study, the animals were killed using an intravenous overdose of narcotics and T61 (Intervet, Unterschleissheim, Germany). Tissue samples (lung, heart, liver, kidney, spleen, and intestine) were immediately harvested and stored in 4% paraformaldehyde at room temperature for 2–3 days. After dehydration, lung tissue specimens were embedded in paraffin for subsequent sectioning. Five micrometer sections were stained with hematoxylin & eosin (H&E) to assess lung tissue morphology. Lung tissue sections were also stained for chloroacetate esterase (CAE) activity to evaluate by bright field microscopy the number of CAE-positive tissue-infiltrating leukocytes per high power field (HPF, 400× magnification). In addition, the wet weight of lung tissues was measured, followed by drying for 10 days in an oven at 70°C for the subsequent calculation of lung wet-to-dry weight ratio.

For microbiological analysis of bacterial infection, bronchoalveolar lavage, peritoneal fluid, blood cultures, and lung tissue were processed according to the standard operation procedures of the accredited (DIN EN ISO 15189) Institute of Medical Microbiology, Virology, and Hygiene, Rostock University Medical Centre. Microbiological sample processing was realized as follows: Gram staining of bronchoalveolar lavage was performed with an automated Gram stainer (Previ Color Gram, bioMérieux, Marcy l’Etoile, France). Specimens were plated on Columbia agar supplemented with 5% sheep blood (BD, Heidelberg, Germany), MacConkey agar (BD), and Chocolate agar (BD), respectively. Agar plates were subsequently cultured at 37°C under a 20% O2/5% CO2 atmosphere for 48 hours. Gram staining of peritoneal fluid and lung tissue was performed with an automated Gram stainer (bioMérieux). Specimens were plated on Columbia agar supplemented with 5% sheep blood (BD), MacConkey agar (BD), Schaedler agar (BD), and SKV agar (BD) in case of peritoneal fluid, on Columbia agar supplemented with 5% sheep blood (BD), MacConkey agar (BD), Chocolate agar (aerobic conditions; BD), Chocolate agar (anaerobic conditions; BD), and SKV agar (BD) in case of lung tissue, respectively. Furthermore, brain heart infusion (BD) and thioglycolate broth were inoculated. Agar plates were subsequently cultured at 37°C under aerobic conditions (20% O2/5% CO2 atmosphere) for 48 hours, and for 120 hours under anaerobic conditions (80% N2/10% CO2/10% H2 atmosphere), broth media at 37°C under ambient atmosphere. Positive broth media (i.e., macroscopically visible turbidity) were subcultured on Columbia and Schaedler agar as described above. Blood cultures (8–10 ml) of venous blood were used to inoculate aerobic and anaerobic blood cultures of the BacT/ALERT 3D system (bioMérieux). Blood cultures were incubated for up to seven days. Positive blood cultures were subcultured on Columbia agar supplemented with 5% sheep blood (BD) and Schaedler agar (BD). Recovered microorganisms were identified by matrix-assisted laser-desorption-ionization time-of-flight mass spectrometry (MALDI-TOF-MS) using a Shimadzu “AXIMA Assurance” MALDI-TOF mass spectrometer (Shimadzu Germany Ltd., Duisburg, Germany). For MALDI-TOF analyses, isolates were prepared using alpha-cyano-4-hydroxy cinnamic acid (bioMérieux) as matrix. Spectral fingerprints were analyzed by using Vitek MS IVD V2, database MS-CE version CLI 2.0.0 (bioMérieux).

Study Design

Of the 15 animals, six animals were treated with the commercial iLA-activve system, seven animals with the I-lung prototype, and two animals served as sham control animals. Unless stated otherwise, all experiments were carried out for 72 hours. Hypercapnia with a target PaCO2 of 10 kPa was generated by reducing the respiratory rate to 6/min and the tidal volume to 4–5 ml/kg BW. Then both extracorporeal devices were switched on for CO2 elimination, using a constant blood flow of 1.2 L/min and an adjustable sweep gas flow, which was increased in steps of 2 L/min for 10 min each to a final value of 8 L/min. The target value was maintained at a PaCO2 of 5–6 kPa.

To repeatedly confirm the respiration-associated hypoventilation and the effective CO2 removal in both systems, extracorporeal decarboxylation was stopped every 12 hours by reducing sweep gas flow to 0 L/min for approximately 45 min to initiate an increase of the PaCO2 toward the initial hypercapnic state with a target PaCO2 value of 10 kPa.

Sham-operated pigs (n = 2) underwent the identical anesthetic procedures and surgical instrumentation, except that no extracorporeal device was connected. Mechanical ventilation was carried out with a respiratory rate of 16/min and a tidal volume of 4–5 ml/kg BW for maintenance of physiologic blood gas parameters and balanced acid base status for more than 72 hours. To allow comparison to the other two groups, the respiratory rate in the sham-operated animals was reduced to 8/min every 12 hours to provoke transient episodes of hypercapnia.

Statistical Analysis

Statistical calculations were performed using SigmaPlot 12.0 (Systat Software Inc., Richmond, CA). Inner-group and inter-group comparisons were only performed between the iLA-activve and I-lung group of pigs. The two sham pigs served to assess the effect of decarboxylation on other potentially confounding factors, such as surgery, anesthesia, and extracorporeal circulation. Significance within the two groups was tested with repeated-measures ANOVA for normal distributed values or with repeated-measures ANOVA on ranks for nonparametric data followed by post-hoc analysis with the Holm-Sidak or Dunn’s method. Significance between the two groups was tested using the t-test for normal distributed values or the Mann–Whitney Rank Sum test for nonparametric data. Results are presented as means ± standard deviation (SD) or as median and 25th–75th percentiles, where appropriate. A p value <0.05 was considered statistically significant.


Fourteen of 15 animals survived until the end of the study. One pig developed septic complications because of severe pneumonia and had to be euthanized after 56 hours. This pig was included in the data analysis up to the 48th hour. For both devices, no technical problems were observed over the observation period of 72 hours. In addition, thromboembolic events and hemolysis were absent in either of the two study groups.

CO2 removal with the prototype I-lung device was as effective as with the established iLA-activve system with a blood flow of 1.2 L/min and a sweep gas flow of 8 L/min (Figure 2). For both systems, PaCO2 values were in the range of 5–6 kPa throughout the 72 hour observation period with no marked differences between the two devices and the two sham-operated pigs (Figure 2). To confirm the respiration-associated hypoventilation and the effective decarboxylation throughout the experiment, the sweep gas flow was reduced repeatedly to 0 L/min to initiate an increase of PaCO2 (Figure 2, grey arrows). By doing so, PaCO2 rapidly increased to the initial hypercapnic values of approximately 10 kPa. Conversely, upon restarting sweep gas flow (8 L/min), PaCO2 dropped to normocapnic values. Taken together, these results indicate that 1) the predetermined parameters of hypoventilation, i.e., a respiratory rate of 6/min and a tidal volume of 4–5 ml/kg BW, were sufficient to induce hypercapnia and that 2) both devices similarly and adequately eliminated CO2, as required for extracorporeal membrane ventilators.

Figure 2.:
PaCO2 more than 72 hours from pigs treated with extracorporeal lung assist devices (black circle) iLA-activve and (white square) prototype I-lung (experimental groups) and from (grey triangle) sham-operated pigs, respectively. After 8 hours and then every 12 hours, sweep gas flow was stopped for 45 min (0 L/min) to provoke an increase PaCO2, as indicated by the arrows. Data are presented as mean ± SD. No statistically significant differences were found either within the individual groups or between groups.

Table 1 presents several key systemic hemodynamic parameters. In general, for both groups with the different extracorporeal membrane ventilators, the experimental animals showed comparable cardiovascular and hemodynamic stability. MAP remained constant over time in both groups. Of note, however, animals of both groups exhibited severe tachycardia at the onset of the study (0 hour: p < 0.05 versus all other time points), with a subsequent decrease of the HRs to physiologic levels. In parallel, sham-operated pigs also showed initially higher HRs (see Table S1, Supplemental Digital Content,, although this was less pronounced than in the groups connected to the iLA-activve and I-lung, respectively. Based on the increased HRs at 0 hour, the initial CO was higher than at later time points in all groups studied. This hemodynamic situation immediately after completion of all interventional steps might simply reflect anesthesia- and surgery-related stress. A negligibly lower FiO2 values at the I-lung animals to the iLA-activve was observed during the study. Statistically significant differences between the results obtained with the two devices were observed after 24 hours and after 48 hours. In addition, FiO2 values show statistically significant inner-group differences (0 vs. 72 hours: p < 0.05).

Table 1.:
Systemic Hemodynamic Parameters more than 72 Hours from Pigs Treated with Extracorporeal Lung Assist Devices iLA-Activve and I-Lung, Respectively

The PaO2 values and blood gas-related parameters did not differ between the iLA-activve and I-lung groups (Table 2): pH, hemoglobin, and electrolytes were stable over time without any inner-group differences. By contrast, metabolic parameters, such as glucose and lactate, were significantly increased at 0 hour when compared with all other time points (p < 0.05), echoing the above-mentioned initial stress condition. In line with this, the initially elevated glucose and lactate levels decreased subsequently. In the sham-operated animals, PaO2 values and blood gas parameters remained stable over time, except of the lactate levels (see Table S2, Supplemental Digital Content, There were no statistically significant inter-group differences.

Table 2.:
PaO2 Values and Blood Gas-Related Parameters more than 72 Hours from Pigs Treated with Extracorporeal Lung Assist Devices iLA-Activve and I-Lung, Respectively

Figures 3 and 5 represent coagulation and hemolysis parameters. Specifically, platelets counts (Figure 3A) and activated partial thromboplastin time (aPTT; Figure 3B) decreased over time in both the iLA-activve and the I-lung group to a comparable extent, suggesting activation of the coagulation cascade and inflammatory response due to membrane contact. In contrast, values of ACT remained almost constant in the range of 150–250 seconds (Figure 4). Hemolysis parameters, such as haptoglobin (Figure 5A) and LDH (Table 3) increased with time. In addition, fibrinogen (Figure 5B) as well as AST and creatinine kinase (Table 3) increased with time. These changes can be interpreted globally as being due to side effects commonly associated with extracorporeal circulation, such as coagulation activation and hemolysis. This conclusion is supported by the fact that sham-operated animals did not show platelet consumption and reduced aPTT (Figure 3A and B).

Table 3.:
Arterial Blood Cell Count and Blood Chemistry Parameters Including Cytokines and sP-Selectin more than 72 Hours from Pigs Treated with Extracorporeal Lung Assist Devices iLA-Activve and I-Lung, Respectively
Figure 3.:
Kinetics of (A) platelet counts and (B) activated partial thromboplastin times in pigs treated for 72 hours with extracorporeal lung assist devices (black box plot) iLA-activve and (white box plot) I-lung and in (grey box plot) sham-operated pigs, respectively. Statistical analysis for inner-group and inter-group comparison was only performed in the iLA-activve and I-lung group of pigs. The two sham-operated pigs were used to discriminate the effect of decarboxylation versus other confounding factors, such as surgery, anesthesia, and extracorporeal circulation. Values are presented as median and 25th–75th percentiles. *p < 0.05 vs. 0 hour, # p < 0.05 vs. iLA-activve.
Figure 4.:
Kinetics of activated clotting time in pigs treated for 72 hours with the extracorporeal lung assist devices (black circle) iLA-activve and (white square) prototype I-lung, respectively. ACT values are presented as mean ± SD. No statistically significant differences were found either within the individual groups or between groups. Horizontal grey lines indicate the range of values of the two sham-operated pigs.
Figure 5.:
Expression levels of hemolysis markers such as (A) haptoglobin and (B) fibrinogen in pigs treated for 72 hours with extracorporeal lung assist devices (black box plot) iLA-activve and (white box plot) I-lung and in (grey box plot) sham-operated pigs, respectively. Statistical analysis for inner-group and inter-group comparison was only performed in the iLA-activve and I-lung group of pigs. The two sham-operated pigs were used to discriminate the effect of decarboxylation versus other confounding factors, such as surgery, anesthesia, and extracorporeal circulation. Values are presented as median and 25th–75th percentiles. No statistically significant differences were found between the groups. *p < 0.05 vs. 0 hour.

Table 3 presents values for arterial blood cell counts and blood chemistry including cytokines and plasma sP-selectin, as well as transaminases and creatinine kinase over 72 hours. AST increased over the time with statistically significant inner-group differences (iLA: 0 vs. 48 hours, 72 hours; I-lung: 0 vs. 48 hours). All other parameters did not show statistically significant inter- and inner-group differences. Cytokine levels in the sham-operated animals did not show any noticeable elevation (see Table S3, Supplemental Digital Content,

The number of CAE-positive cells in lung tissue (median, 25th–75th percentiles) was slightly higher in the iLA-activve group (20.8 cells/HPF, 11.6–53.1) and in the I-lung group (15.3 cells/HPF, 10.6–30.1) when compared with the two sham-operated pigs (9.9 and 15.5 cells/HPF).

No differences in the wet-to-dry ratio of lung tissue were recorded between the study groups (iLA-activve: 34.8 ± 18.6 g; I-lung: 31.8 ± 16.5 g) and the two sham-operated pigs (31.9 and 31.9 g).

Macroscopically and microscopically analyses of bronchoalveolar lavage, peritoneal fluid, and lung tissue yielded no signs of infection, confirmed by microbiological analyses, mainly displaying bacterial flora (see Table S4, Supplemental Digital Content, Blood cultures were negative for all pigs except for one, revealing Salmonella enterica ssp. enterica ser. Derby.


In this study, we tested, for the first time, the new prototype of a miniaturized and wearable membrane ventilator system (I-lung) in a large animal model and compared its performance to that of the well-established iLA-activve extracorporeal membrane ventilator. For all parameters tested, the performance of the new system was comparable to that of the established system. Actually, when adjusted to the surface area of gas exchange membranes, comparative analysis of CO2 elimination in both devices revealed a nearly two times higher capacity of I-lung (216 ml/min per m2 surface) versus iLA-activve (114 ml/min per m2 surface). The more than 50% smaller gas exchange surface area of the I-lung (0.6 m2) and that of the iLA-activve (1.3 m2) yield almost identical CO2 values at the outlet of the gas exchanger, suggesting that the efficiency of gas exchange of the novel I-lung is nearly double that of clinically approved iLA-acctive. The reason for this increase in efficiency is currently under investigation and may be related to the altered geometry and improved flow parameters in the I-lung.

The main result of this study is that both pump-driven veno-venous lung assist devices reliably eliminate CO2 from the animal’s blood over the entire 72 hour study. A 72 hour experiment is challenging from several aspects, i.e., maintenance of anesthesia, global hemodynamics, and metabolism, as well as homeostasis of coagulation and inflammation cascades. To the best of our knowledge, this is the first time that extracorporeal membrane ventilators were tested in such a comprehensive fashion in a long-term set-up. In previous studies such membrane ventilators had been tested preclinically only for up to 48 hours.6–8

To confirm the efficacy of both devices over time, airflow was stopped briefly every 12 hours and hypercapnic ventilatory status was reestablished. Upon restarting sweep gas flow, PaCO2 values rapidly returned to physiologic values. These interventions suggest that 1) the predetermined parameters of hypoventilation were able to efficiently induce life-threatening hypercapnia, and 2) that both devices adequately and equally eliminate CO2 over the time period studied. In addition to an excellent CO2 elimination, a relevant oxygenation in both devices certainly takes place. Because of the 50% smaller membrane surface and an inferentially higher specific blood flow in the new devices I-lung, the lower FiO2 values could be justified.

As a caveat, pigs with a mean BW of approximately 50 kg, as used in the current study, might not fully mimic the situation in adult humans in terms of cannulas and flow rates meeting the higher demand for CO and metabolism. In contrast to other porcine studies, in which smaller catheter calibers for extracorporeal lung assist devices were used,6–8 the 22 Fr double-lumen catheters used in the current study for a single venovenous exchange allow for increasing device performance to levels similar to what would be necessary in humans. This is confirmed by a recent study in critical ill patients, which used a double-lumen cannula with 22 Fr for a single jugular or 24 Fr for single femoral access.4

The safe use of extracorporeal membrane ventilators is markedly jeopardized by the imbalance of plasma coagulation, in particular during long-term application. Patients on extracorporeal life support need to be carefully anticoagulated to maintain the patency of the blood circulation without causing internal bleeding or avoiding thrombosis.9 In accordance with clinical practice, ACT values between 180 and 220 seconds10 can be maintained by adjusting the rate of heparin infusion to between 20 and 50 U/kg//h.9 In the current study, ACT values were targeted to ca. 200 seconds, which required infusion of heparin in the range of 20 and 32 U/kg/h, a dosage that correlates well with the recently reported data.9 Under this regime, we observed neither hemolysis nor thrombosis during the 72 hour use of either of the heparin-coated devices, I-lung and iLA-activve. Beside stable ACT values, the international normalized ratio (INR) values were also constant in both groups of instrumented animals. There seems to be an obvious discrepancy between the stable ACT and INR values and the decrease in aPTT. However, ACT does not directly correlate with either aPTT or heparin levels.9 In addition, aPTT determination has several limitations, such as reproducibility and variability.9

Despite the absence of overt thromboembolic events, a more detailed analysis of blood parameters revealed a progressive decline in platelet count and aPTT as well as a steady increase of fibrinogen, suggesting some activation of the coagulation system. In line with this observation, Kopp et al.7, analyzed the performance of a nonheparin coated, miniaturized extracorporeal membrane ventilator with integrated rotary blood pump in pigs for a total of 4 hours and reported increased levels of thrombin-antithrombin III complex and decreased platelet counts, although the ACT levels were in the target range. Importantly, however, in our study the miniaturized I-lung prototype and the clinically approved iLA-activve system did not differ in terms of activation of the coagulation cascade for more than 72 hours and there were no clinical manifestations of overt thromboembolic complications.

In general, the pigs used in our study were hemodynamically stable during the whole observation period, although all animals, including the sham-operated ones initially, after completion of all interventional steps, revealed signs of anesthesia- and surgery-related initial stress, such as increased HR and lactate with subsequent return to physiologic values. In line with this observation, Zanella et al.6 reported the transient elevation of lactate levels after complex and invasive interventions in a pig model of low-flow venovenous extracorporeal ventilation. In this context, lactate clearance is not impaired in patients with severe sepsis and cardiogenic shock.11

Although novel and original, our study has also several limitations: First, the supine position of the pigs on the table was not changed over time, as is routinely performed in patients to prevent atelectasis.12 Thus, posture-related artifacts in lung histology as well as atelectasis-induced inflammatory processes cannot be completely discounted. One could only speculate on the isolation of Salmonella Derby from a blood culture of one animal. Possible reasons could be a bacteremia posttranslocation from the intestines, septicemia, or simple contamination during sampling. Because the animal did not display severe signs of systemic infection, the latter explanation is probably the most plausible. Although no adverse effects were detected histologically, the slight increase of cytokines toward the end of the observation period may cause the onset of a systemic inflammatory response. As the sham-operated animals showed very low cytokine levels, an influence of the surgical procedure on the inflammatory mediator release can probably be excluded. Thus, the trend to an elevated release of cytokines in the iLA-activve and I-lung groups, although moderate, might be attributed to the extracorporeal devices. Second, hypercapnia was artificially induced by hypoventilation in healthy pigs. This model does not definitely mirror the pulmonary disorders seen in COPD patients. In future studies, we will more faithfully simulate the clinical situation by employing an acute lung failure model. However, as a drawback, such a model would lack standardized hypercapnia conditions with constant high levels of PaCO2.


We established a preclinical in vivo porcine model for comparative testing of two pump-driven venovenous extracorporeal lung assist devices, i.e., the well-established iLA-activve versus the first prototype of I-lung, a new miniaturized, wearable device. In this first study, the prototype proved to be a safe, feasible, and efficient device for adequate decarboxylation without any adverse events, comparable in its efficiency and safety to the commercially available system. Once approved for clinical use, the new miniaturized, wearable, and transportable device represents a promising new tool for treatment of awake and mobilized patients with decompensated pulmonary disorders.


The authors express their thanks to Sven Hamberger (engineer at Novalung GmgH, Heilbronn, Germany) for his continuous technical support during the experiments. The authors gratefully acknowledge funding by the EU (FP7-HEALTH Program, Project No. 304932).


1. Decramer M, Janssens W, Miravitlles M. Chronic obstructive pulmonary disease. Lancet. 2012.379: 13411351.
2. Sklar MC, Beloncle F, Katsios CM, Brochard L, Friedrich JO. Extracorporeal carbon dioxide removal in patients with chronic obstructive pulmonary disease: A systematic review. Intensive Care Med. 2015.41: 17521762.
3. Abrams D, Roncon-Albuquerque R Jr, Brodie D. What’s new in extracorporeal carbon dioxide removal for COPD? Intensive Care Med. 2015.41: 906908.
4. Hermann A, Staudinger T, Bojic A, et al. First experience with a new miniaturized pump-driven venovenous extracorporeal CO2 removal system (iLA Activve): A retrospective data analysis. ASAIO J. 2014.60: 342347.
5. Hermann A, Riss K, Schellongowski P, et al. A novel pump-driven veno-venous gas exchange system during extracorporeal CO2-removal. Intensive Care Med. 2015.41: 17731780.
6. Zanella A, Mangili P, Redaelli S, et al. Regional blood acidification enhances extracorporeal carbon dioxide removal: A 48-hour animal study. Anesthesiology. 2014.120: 416424.
7. Kopp R, Bensberg R, Arens J, et al. A miniaturized extracorporeal membrane oxygenator with integrated rotary blood pump: Preclinical in vivo testing. ASAIO J. 2011.57: 158163.
8. Kopp R, Bensberg R, Wardeh M, Rossaint R, Kuhlen R, Henzler D. Pumpless arterio-venous extracorporeal lung assist compared with veno-venous extracorporeal membrane oxygenation during experimental lung injury. Br J Anaesth. 2012.108: 745753.
9. Annich GM. Extracorporeal life support: The precarious balance of hemostasis. J Thromb Haemost. 2015.13(suppl 1): S336S342.
10. Murphy DA, Hockings LE, Andrews RK, et al. Extracorporeal membrane oxygenation-hemostatic complications. Transfus Med Rev. 2015.29: 90101.
11. Revelly JP, Tappy L, Martinez A, et al. Lactate and glucose metabolism in severe sepsis and cardiogenic shock. Crit Care Med. 2005.33: 22352240.
12. Kredel M, Bischof L, Wurmb TE, Roewer N, Muellenbach RM. Combination of positioning therapy and venovenous extracorporeal membrane oxygenation in ARDS patients. Perfusion. 2014.29: 171177.

decarboxylation; extracorporeal membrane ventilator; porcine model; respiratory failure

Supplemental Digital Content

Copyright © 2017 by the ASAIO