Share this article on:

Experimental Venoarterial Extracorporeal Membrane Oxygenation Induces Left Ventricular Dysfunction

Schiller, Petter; Vikholm, Per; Hellgren, Laila

doi: 10.1097/MAT.0000000000000392
Adult Circulatory Support

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) has experienced an increased use in acute cardiac failure. There are some reports on negative effects of VA-ECMO on cardiac function, such as left ventricular (LV) dilatation and cardiac stun, but the support in the literature is scarce. This study investigates the effects of experimental VA-ECMO on LV function in both peripheral and central cannulation. Ten pigs were randomized to VA-ECMO by either peripheral cannulation through the femoral vessels or central cannulation in the right atrium and ascending aorta. Left ventricular performance was measured with pressure-volume catheters during 5 hours of VA-ECMO. The LV end-diastolic and end-systolic volumes increased comparably in both groups during ECMO. Left ventricular ejection fraction, stroke work, and maximum rate of pressure change decreased comparably in both groups as a function of time on ECMO. The site of cannulation had no impact on the LV response to ECMO. In conclusion, VA-ECMO increased LV volumes and reduced LV function, irrespective of cannulation site in this experimental model. Reduced LV ejection fraction and stroke work indicated LV dysfunction during ECMO.

From the Department of Cardiothoracic Surgery, Uppsala University Hospital, Uppsala, Sweden.

Submitted for consideration August 2015; accepted for publication in revised form May 2016.

Disclosures: The authors have no conflicts of interest to report.

The research was supported by grants from the Uppsala University Hospital.

Correspondence: Laila Hellgren, MD, PhD, Department of Cardiothoracic Surgery, Uppsala University Hospital, SE-75185, Uppsala, Sweden. Email:

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) has experienced an increased use in patients with cardiac failure and is nowadays an established treatment option in acute heart failure1 and as an adjunct to conventional cardiopulmonary resuscitation (CPR), known as extracorporeal CPR.2 The primary aim of VA-ECMO in the adult cardiac patient is to restore circulation, ensure organ perfusion and provide left ventricular (LV) unloading to allow for myocardial recovery. However, some concerns are raised regarding the effects of VA-ECMO on cardiac function3,4—and negative effects on LV function during ECMO have been reported. These negative effects were also reported from patients with normal cardiac function when VA-ECMO was instituted because of pulmonary failure. The reported negative effects include cardiac stun,5 dilated LV, and pulmonary edema6,7; even electromechanical dissociation has been experienced.8

In addition, another important aspect of VA-ECMO is the site of cannulation. The site varies, but in most centers, peripheral cannulation through the femoral vessels is a standard strategy for ECMO initiation. Central cannulation is most often used when postcardiotomy patients are in need for support. It is discussed whether the antegrade flow in the aorta produced by central arterial cannulation would have a positive effect on LV unloading and thereby LV function, and some centers even advocate conversion to central cannulation in case of LV dysfunction; although there are no reliable evidence to support this strategy.

Given the lack of research investigating LV response to VA-ECMO, we conducted an experimental study to investigate the effects on LV function during VA-ECMO. To diminish the impact of cannulation site, randomization was performed to either peripheral or central cannulation. The primary aim of the study was to establish the LV response to VA-ECMO measured by conductance catheter and pressure-volume loops.

Back to Top | Article Outline

Materials and Methods

Uppsala University Ethical Committee on Laboratory Animal Research approved the study. All animals received humane care in compliance with Swedish legislation on animal experimentation (Animal Welfare Act SFS1998: 56). The study comprised 10 pigs of Swedish country breed with a mean weight of 38 kg (range, 33–47 kg). The animals were randomized to peripheral cannulation (5 pigs) or central cannulation (5 pigs). The randomization was performed by drawing lots.

Back to Top | Article Outline


Anesthesia was induced and maintained according to our previously published protocol.9 The animals were intubated and mechanically ventilated with 35% oxygen. Volume-controlled ventilation was used aiming for an arterial pCO2 within the range of 5.0–5.5 kPa, and positive end-expiratory pressure of 5 cm H2O was applied. The bladder was catheterized with a silicone catheter, and body temperature was controlled with a heating pad.

Two catheters were inserted through the external jugular vein (BD Careflow 17G; Becton-Dickinson AB, Stockholm, Sweden); one for central venous pressure, blood sampling, and drug administration. The second was advanced into the right ventricle (RV) for pressure measurements. A pulmonary artery (PA) catheter (BD Criticath Pulmonary Artery/Thermodilution Catheter; Becton-Dickinson AB) was introduced through the right jugular vein into the PA. For monitoring of the arterial blood pressure, a catheter was inserted through the right femoral artery into the descending aorta (BD Careflow 20G; Becton-Dickinson AB). A 5 French 12 electrode pressure-volume (PV) conductance catheter (Ventri-Cath, Millar Instruments, Oxford, UK) was introduced through the left carotid artery, then through the aortic valve, and placed in the LV. The catheter was connected to the Millar MPVS Ultra analyzing system (Millar Instruments). Catheter positioning was determined by monitoring individual segmental PV loops. Extraventricular segments were excluded from analysis.

Back to Top | Article Outline

Surgical Preparation

For peripheral cannulation, the femoral vessels in the left groin were dissected free without entering the abdominal cavity. After administration of one dose of 7,500 IU heparin, the femoral artery was cannulated with a 16 French artery cannula (FemFlex; Edwards Lifesciences Nordic AB, Malmö, Sweden). The tip of the cannula was positioned near the aortic bifurcation. For distal perfusion, a small cannula (BD Careflow 17 G; Becton-Dickinson AB) was introduced distally into the femoral artery and connected to the main arterial cannula. A 19 French venous cannula (Bio-Medicus Venous Cannula; Medtronic Inc., Minneapolis, MN) was introduced through the femoral vein and advanced with the tip located in the right atrium. For central cannulation, a midline sternotomy was performed. After one dose of 7,500 IU heparin, a 16 French arterial cannula (FemFlex; Edwards Lifesciences Nordic AB) was introduced 3 cm into the ascending aorta, directed toward the descending aorta. The right atrium was cannulated with a 28 French venous cannula (Malleable Single Stage Venous Cannula, Edwards Lifesciences Nordic AB).

Arterial and venous cannulae were deaired and connected to an ECMO system comprising an adult membrane oxygenator (Quadrox D, Maquet Critical Care AB, Solna, Sweden) and a Bio-Pump (Medtronic Inc.). The oxygenator was connected to a heater-cooler for maintaining the normal body temperature.

Back to Top | Article Outline

Experimental Protocol

After baseline measurements, peripheral or central VA-ECMO was initiated at a blood flow of 100 ml/kg/min. Sweep gas through the oxygenator was set to FiO2 1.0 and the gas flow was adjusted to achieve normo-ventilation. Heparin was administered with the dosage of 5,000 IU every hour on ECMO. Venoarterial ECMO was continued for 5 consecutive hours (Figure 1). At the end of the experiment, the animals were sacrificed by intravenous administration of 80 mmol potassium chloride. Postmortem examination was performed to verify correct positioning of the cannulae.

Back to Top | Article Outline


Electrocardiogram (ECG), systolic arterial blood pressure (SAP), mean arterial blood pressure, diastolic arterial blood pressure (DAP), central venous pressure (CVP), RV pressure (RVP), mean PA pressure (mPAP), and pulmonary capillary wedge pressure (PCWP) were monitored and recorded. Cardiac output (CO) was measured using the thermodilution technique at baseline. During ECMO support, the LV CO was registered with the conductance catheter. The PV catheter was calibrated for blood resistivity using a rho cuvette (Millar Instruments), and the alpha-value was calculated using thermodilution-derived CO. Pressure-volume loops from the LV were recorded for 25 consecutive heartbeats with the ventilation suspended at end-expiration. Left ventricular systolic pressure (LVPsys), LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV stroke volume (LVSV), LV stroke work (LVSW), LV ejection fraction (LVEF), and maximum rate of LVP change (LV dP/dtmax) were extracted from the PV loops. Hemodynamic measurements and PV loops were recorded every minute for 5 minutes at baseline, after the start of ECMO, and on every hour of ECMO for 5 consecutive hours. Blood samples were collected for blood gas analysis and blood plasma collection. Arterial blood gases, hemoglobin concentration, serum lactate, and mixed venous oxygen saturation (SvO2) were measured with an ABL 500 Radiometer (Medical ApS, Brønshøj, Denmark). All blood tests were analyzed according the routine laboratory protocol at the Department of Clinical Chemistry, Uppsala University Hospital.

Back to Top | Article Outline

Data Analysis and Statistical Methods

For data analysis of the PV loops, the LabChart software (ADInstruments Pty Ltd, Bella Vista, NSW, Australia) was used. As there was no calibration for parallel conductance, baseline LVEF was set to 0.6 and the volumes were calibrated according to this figure.

For CO, CVP, RVP, mPAP, systemic arterial pressure, PCWP, heart rate, LVP, LVSW, LVEF, LVEDV, and LVESV, the value for each pig was calculated as the mean of all measurements in a treatment. The median of all pigs in a treatment was used. All data are presented as medians with interquartile range in brackets. Significance testing was performed using related samples Wilcoxon signed-rank test comparing the values between baseline and ECMO start. Testing for significant changes during ECMO was performed using repeated measures analysis of variance using the Greenhouse-Geisser correction for sphericity. p Values ≤0.05 were considered statistically significant. All statistical calculations were performed in IBM SPSS Statistics software (SPSS Inc, Chicago, IL).

Back to Top | Article Outline


All pigs were successfully cannulated and included in the full study protocol (n = 10). Postmortem examination revealed correct cannula position in all animals, and no structural cardiac anomalies were found.

Back to Top | Article Outline

Effects of VA-ECMO on Left-Sided Pressures

Hemodynamic data are presented in Table 1. During the 5 hours of treatment, there was a constant decrease in SAP from 104 mm Hg (98–113) to 81 mm Hg (70–85) (p < 0.01). In contrast, the DAP increased from 58 mm Hg (53–62) at baseline to 86 mm Hg (78–94) at ECMO start (p < 0.01). Thereafter, during the 5 hours of ECMO, there was a decrease in DAP from 86 mm Hg (78–94) to 67 mm Hg (61–73) at the end of the experiment (p < 0.01). The LVPsys decreased from 95 mm Hg (81–99) at ECMO start to 74 mm Hg (54–83) after 5 hours of ECMO (p < 0.01). Pulmonary capillary wedge pressure decreased from 12 mm Hg (9.0–14) at baseline to 8.0 mm Hg (5.0–11) at ECMO start (p < 0.01), but then remained constant during the 5 hours of ECMO support.

Back to Top | Article Outline

Effects of VA-ECMO on LV Function and Volumes

Left ventricular stroke volume decreased from 44 mm Hg (32–68) to 27 mm Hg (24–38) from baseline to ECMO start. Thereafter, there was no significant change in LVSV during the 5 hour ECMO period. There was a trend toward decreased LV CO during the period on ECMO from 3.7 L/min (2.3–4.9) to 1.5 L/min (0.9–2.2) although this trend did not reach significance. Moreover, the LVSW decreased from 2,968 mm Hg*ml (1,480–3,940) at baseline to 1,661 mm Hg*ml (945–1,944) at ECMO start (p = 0.03). Thereafter, the LVSW gradually decreased during 5 hours of ECMO from 1661 mm Hg*ml (945–1,944) to 608 mm Hg*ml (287–789) at the end of the experiment (p = 0.05 as shown in Figure 2). LVESV increased from 31 ml (28–44) at ECMO start to 55 ml (44–91) (p < 0.01), and LVEDV from 61 ml (48–71) to 84 ml (60–104) (p < 0.01), at the end of the experiment, Figures 3 and 4, respectively.

The combination of a lower LVSV and increased LVEDV resulted in a decrease in LVEF as a function of time on ECMO from baseline at 60% to 26% at the end of support. Left ventricular dP/dtmax decreased gradually from 1,244 mm Hg/sec at baseline to 859 mm Hg/sec (676–1,174) at ECMO start, and finally to 516 mm Hg/sec (368–567) after 5 hours on ECMO (p < 0.01; Figures 5 and 6).

Back to Top | Article Outline

Blood Gases and Biochemical Markers During VA-ECMO

Blood gas analysis revealed increased pO2, SaO2, and SvO2 as ECMO was initiated (Table 2). There was a trend toward lowering lactate values during the ECMO course.

Plasma levels of troponin I, creatinine, ALT, and total protein at different time points during the experiment are shown in Table 3. Plasma analysis revealed a continuous increase in troponin I levels during the experiment from 0.1 µg/L (0.12–0.20) at baseline to 1.65 µg/L (0.84–3.0) at the end of experiment (p < 0.01). There were minor changes in plasma creatinine, ALT, and total plasma protein levels when baseline values were compared with ECMO start, as seen in Table 3, but thereafter there were no changes in these markers.

Back to Top | Article Outline


The main findings of this study were increased LVESV and LVEDV throughout the ECMO treatment, illustrating a distended LV during VA-ECMO. Moreover, LV dP/dtmax, LVSW, and LVEF decreased constantly during ECMO support in both groups, supporting that LV contractility was negatively affected by VA-ECMO.

This is the first study on LV response and hemodynamic changes as measured by PV conductance catheter in both closed chest with peripherally cannulated animals and open chest, during VA-ECMO support. In addition, the study evaluates mode of cannulation and its impact on LV response, and this has not been described previously either. These are two novel aspects of VA-ECMO treatment, which we believe motivate our study.

Previous studies have shown conflicting results on cardiac function during ECMO support. For example, Shen et al.14 reported a significant decrease in load-dependent measures of contractility such as LV function, measured as velocity of circumferential fiber length shortening (VCF) and LV shortening fraction (LVSF) in their model of VA-ECMO for 4 and 6 hours in six centrally cannulated, open chest, pigs. In addition, in their study, LV dP/dt had a decreasing trend and that is consistent with our findings. Moreover, Pyles et al.15 conducted a study on peripheral VA-ECMO in newborn lambs with or without induced hypoxemia. In the normally oxygenated group, they noted a similar decrease in LV function measured as VCF, LVSF, and LV dP/dt after 5 minutes or 1 hour on ECMO, indicating that LV performance is affected negatively by VA-ECMO. Another study by Tanke et al.16 investigated cardiac function in 29 neonatal patients with or without persistent ductus arteriosus (PDA) on peripheral VA-ECMO for respiratory failure. In the group without PDA, they noted increased LV dimensions and reduced LV fractional shortening after 24 to 36 hours on ECMO, supporting our experimental findings. The support for the negative impact of V-A ECMO on LV function is accumulating, and our study adds to this believe. However, if these effects have clinical relevance during the period of treatment or negative long-term effect, affecting outcome is still uncertain. At the other side of the spectrum, VA-ECMO is today established treatment with survival benefit for acute cardiogenic shock.

Several ECMO centers have experienced good clinical outcome, such as reversal of pulmonary edema, when volume unloading of the left ventricle is to the conventional VA-ECMO.17 However, there is no support in the literature for more favorable outcome related to unloading the LV during VA-ECMO. This approach has evolved in clinical practice when there are problems with mainly pulmonary edema during support. Left ventricular volume unloading can be achieved in several ways, including left atrial or ventricular vent insertion,6, 18–20 atrial septostomy,21, 22 intra-aortic balloon counterpulsation, or by addition of a LV assist device such as the Impella (Abiomed Inc, Danvers, MA) or iVAC pump (PulseCath BV, Arnhem, NE).7, 23, 24 Although there are several reports on positive clinical effects of LV unloading during VA-ECMO, these are exclusively in the form of case reports, and it has been suggested that conversion to central cannulation in selected patients on VA-ECMO might facilitate LV unloading. In one recent study on experimental VA-ECMO on swine, the authors claim that both 50% and 100% ECMO support provide some degree of LV unloading shown by decreased LA pressure and LVEDV pressure. However, they did not use conductance measurement of pressure and volume but solely measures of pressure. The usage of conductance catheters for assessment of the relation of pressure and volume must be regarded as superior when exploring a complex matter of loading. In comparison, our study showed increased end-systolic and end-diastolic volumes and reduced stroke work of the LV, indicating that overall LV performance is reduced and some degree of LV distension occurs during VA-ECMO. In addition, it is important to remember that most patients receiving cardiac VA-ECMO always have some degree of LV dysfunction at the time of ECMO initiation, which would probably result in more severe LV dysfunction than observed in this study. In addition, there is a reason to believe that the phenomenon of LV distension would be even more significant in the clinical setting with acute cardiac ECMO. Therefore, unloading the LV in the setting of VA-ECMO seems relevant, at least if there are signs of LV distension. Moreover, to relieve the LV during ECMO support, the reduction of afterload in combination with mechanical LV unloading is of uttermost importance.

As the animals in our study were normally ventilated and left atrial blood oxygenation ought to be normal, the observed changes in contractility should most certainly not be attributed to coronary ischemia as a result of poor oxygenation in the aortic root, which has been suggested as a possible mechanism for LV dysfunction during VA-ECMO.10 Furthermore, there were no ECG signs of coronary ischemia. In a recent experimental study on ECMO and coronary perfusion, a significant increase in LAD blood flow was seen during 100% ECMO support. They showed that the LAD blood flow was significantly higher than during the unsupported failing heart, thereby supporting myocardial recovery. In our study, there was, however, a gradual increase in troponin I in both groups, suggesting that a structural damage to the myocardium occurred. Although the main clinical application for cardiac troponins is in the diagnosis of myocardial ischemia, several studies have shown a rise in cardiac troponins in nonischemic heart failure where the underlying reason might be associated with changes of pressure and load.11–13 Moreover, local and circulating inflammatory markers including tumor necrosis factor α, interleukin 6, and reactive oxygen species may lead to direct myocardial injury by cytotoxic effects, and it is well established that extracorporeal circulation causes a large inflammatory response. Although the underlying mechanism for the increased levels of troponins is somewhat unclear, these results support the negative effects of VA-ECMO on LV function. Finally, central cannulation showed no superior effects compared with peripheral cannulation in any of the measured variables.

Back to Top | Article Outline


This study has several limitations. First, this is an experimental study with a small sample size, and therefore, detection of differences regarding cannulation site may be difficult. Also, healthy animals were placed on VA-ECMO, and the observed changes during ECMO support may differ considerably in patients who receive ECMO with complex cardiopulmonary disorders. The use of the golden standard method echocardiography for cardiac performance measurements was not a reliable option in our study because both the technique and the measurements will differ considerably when an open chest is compared with a closed chest. Although the period of support might be regarded as brief, experimental ECMO of most studies is even shorter. Also, the evolution of LV dysfunction is not possible to establish during 5+ hours of support, for this end point longer studies are needed.

Back to Top | Article Outline


Left ventricular function was progressively and negatively affected during the ECMO treatment. Moreover, LV volumes increased and cardiac markers show signs of negative effect on the myocardium. This study implies that the use of LV unloading during VA-ECMO might be motivated to reduce accumulated LV volumes, although the support for the actual beneficial effects of LV drainage is not yet established.

Back to Top | Article Outline


1. Matsumiya G, Saitoh S, Sawa Y: Extracorporeal assist circulation for heart failure. Circ J 2009.73: A42–A47.
2. Huang SC, Wu ET, Chen YS, et al: Extracorporeal membrane oxygenation rescue for cardiopulmonary resuscitation in pediatric patients. Crit Care Med 2008.36: 1607–1613.
3. Bavaria JE, Ratcliffe MB, Gupta KB, Wenger RK, Bogen DK, Edmunds LH Jr: Changes in left ventricular systolic wall stress during biventricular circulatory assistance. Ann Thorac Surg 1988.45: 526–532.
4. Walther FJ, van de Bor M, Gangitano ES, Snyder JR: Left and right ventricular output in newborn infants undergoing extracorporeal membrane oxygenation. Crit Care Med 1990.18: 148–151.
5. Martin GR, Short BL, Abbott C, O’Brien AM: Cardiac stun in infants undergoing extracorporeal membrane oxygenation. J Thorac Cardiovasc Surg 1991.101: 607–611.
6. Swartz MF, Smith F, Byrum CJ, Alfieris GM: Transseptal catheter decompression of the left ventricle during extracorporeal membrane oxygenation. Pediatr Cardiol 2012.33: 185–187.
7. Koeckert MS, Jorde UP, Naka Y, Moses JW, Takayama H: Impella LP 2.5 for left ventricular unloading during venoarterial extracorporeal membrane oxygenation support. J Card Surg 2011.26: 666–668.
8. Rosenberg EM, Cook LN: Electromechanical dissociation in newborns treated with extracorporeal membrane oxygenation: An extreme form of cardiac stun syndrome. Crit Care Med 1991.19: 780–784.
9. Vikholm P, Schiller P, Johansson J, Hellgren L: A modified Glenn shunt improves haemodynamics in acute right ventricular failure in an experimental model. Eur J Cardiothorac Surg 2013.43: 612–618.
10. Nakamura T, Takata M, Arai M, Nakagawa S, Miyasaka K: The effect of left-to-right shunting on coronary oxygenation during extracorporeal membrane oxygenation. J Pediatr Surg 1999.34: 981–985.
11. Bouhemad B, Nicolas-Robin A, Arbelot C, Arthaud M, Féger F, Rouby JJ: Acute left ventricular dilatation and shock-induced myocardial dysfunction. Crit Care Med 2009.37: 441–447.
12. Felker GM, Hasselblad V, Tang WH, et al: Troponin I in acute decompensated heart failure: Insights from the ASCEND-HF study. Eur J Heart Fail 2012.14: 1257–1264.
13. Bass A, Patterson JH, Adams KF Jr: Perspective on the clinical application of troponin in heart failure and states of cardiac injury. Heart Fail Rev 2010.15: 305–317.
14. Shen I, Levy FH, Vocelka CR, et al.: Effect of extracorporeal membrane oxygenation on left ventricular function of swine. Ann Thorac Surg. 2001.71: 862–867.
15. Pyles LA, Gustafson RA, Fortney J, Einzig S: Extracorporeal membrane oxygenation induced cardiac dysfunction in newborn lambs. J Cardiovasc Transl Res 2010.3: 625–634.
16. Tanke RB, Daniëls O, van Heijst AF, van Lier H, Festen C: Cardiac dimensions during extracorporeal membrane oxygenation. Cardiol Young 2005.15: 373–378.
17. Fuhrman BP, Hernan LJ, Rotta AT, Heard CM, Rosenkranz ER: Pathophysiology of cardiac extracorporeal membrane oxygenation. Artif Organs 1999.23: 966–969.
18. Aiyagari RM, Rocchini AP, Remenapp RT, Graziano JN: Decompression of the left atrium during extracorporeal membrane oxygenation using a transseptal cannula incorporated into the circuit. Crit Care Med 2006.34: 2603–2606.
19. Barbone A, Malvindi PG, Ferrara P, Tarelli G: Left ventricle unloading by percutaneous pigtail during extracorporeal membrane oxygenation. Interact Cardiovasc Thorac Surg 2011.13: 293–295.
20. Avalli L, Maggioni E, Sangalli F, Favini G, Formica F, Fumagalli R: Percutaneous left-heart decompression during extracorporeal membrane oxygenation: an alternative to surgical and transeptal venting in adult patients. ASAIO J 2011.57: 38–40.
21. Pantalos GM, Sahetya S, Merkley TL, Horrell T, Austin EH 3rd, Mascio CE: Mock circulation simulation of extracorporeal membrane oxygenation support for systemic ventricular failure in an infant: the effect of atrial septostomy. ASAIO J 2012.58: 415–419.
22. Dahdouh Z, Roule V, Lognone T, Sabatier R, Grollier G: Percutaneous blade and balloon atrioseptostomy as a supplement to extracorporeal membrane oxygenation as a bridge to heart transplantation. Cardiovasc Revasc Med. 2012.13: 69–71.
23. Chaparro SV, Badheka A, Marzouka GR, et al: Combined use of Impella left ventricular assist device and extracorporeal membrane oxygenation as a bridge to recovery in fulminant myocarditis. ASAIO J 2012.58: 285–287.
24. Anastasiadis K, Chalvatzoulis O, Antonitsis P, Tossios P, Papakonstantinou C: Left ventricular decompression during peripheral extracorporeal membrane oxygenation support with the use of the novel iVAC pulsatile paracorporeal assist device. Ann Thorac Surg 2011.92: 2257–2259.

extracorporeal membrane oxygenation; cannulation; left ventricle; left ventricular assist device

Copyright © 2016 by the American Society for Artificial Internal Organs