Secondary Logo

Journal Logo

How to do it Article

A Protocol that Mandates Postoxygenator and Arterial Blood Gases to Confirm Brain Death on Venoarterial Extracorporeal Membrane Oxygenation

Ihle, Joshua F.*,†,‡; Burrell, Aidan J. C.*,‡,§; Philpot, Steve J.†,¶; Pilcher, David V.*,‡,§; Murphy, Deirdre J.‡,¶; Pellegrino, Vincent A.*,‡,§

Author Information
doi: 10.1097/MAT.0000000000001086


As the use of veno-arterial extracorporeal membrane oxygenation (V-A ECMO) becomes more widespread, the number of potential brain dead patients on V-A ECMO is also likely to increase.1–3 Bedside clinical assessment of brain death (BD) is recommended even if considering imaging modalities.4 Local and international guidelines state that apnea without signs of ventilatory effort, in the presence of an elevation of the partial pressure of arterial (and therefore by implication cerebral) carbon dioxide (PaCO2) >60 mmHg (or a rise of 20 mmHg above baseline for a chronically hypercapnic patient) and a pH <7.3, are required for the clinical determination of BD.5 The apnea test (AT) must not result in hypoxia or hemodynamic instability. Published guidelines do not describe the process for performing the AT for patients on V-A ECMO.

During peripheral V-A ECMO support, with the recovery of myocardial function, blood may flow to the proximal aorta from the native heart and lungs.6 This can mix with the ECMO circulation at a point in the aorta known as the “water shed” or “mixing point”.7 Different gas tensions can be measured depending on where in the arterial system blood is sampled, relative to the mixing point.8 As the mixing point moves more distally in the aorta, differing gas tensions of oxygen (PaO2) and PaCO2 can result within the arterial system. Thus, the PaCO2 if sampled at only one single point may not reflect the PaCO2 within the cerebral circulation.

The aims of this study were to develop a safe and reliable AT protocol that remains valid in the presence of differing arterial gas tensions for patients on V-A ECMO support, and then to review the impact of the protocol’s implementation on patients in our institution.



The Alfred Hospital in Melbourne, Victoria, Australia is a 45-bed capacity quaternary intensive care unit (ICU), which admits 3,000 patients per year of whom approximately 100 require ECMO, the largest ECMO program in Australia.

Physiologic Considerations

During V-A ECMO regular measurement of the native cardiac output is difficult and it is impossible to reliably determine where the mixing point lies. Consequently, the presence of differing oxygen (O2) and carbon dioxide (CO2) tensions in the arterial system has important implications for sampling of blood during the AT,9 as the brain is supplied by cerebral vessels midway in the arch of the aorta.

The Impact of the “Mixing Point” on Cerebral Blood Flow

In patients with little native blood flow and a proximal aortic mixing point (Figure 1A), right radial arterial blood gases will reflect ECMO blood gas tensions, and this in turn will reflect cerebral gas tensions as ECMO blood supplies the cerebral vessels. However, in patients with recovering myocardial function, the mixing point may move more distally beyond the origin of the right brachiocephalic (and therefore right radial) artery. A right radial arterial blood gas will reflect the native circulation through the heart and lungs, and not the ECMO circuit. In this circumstance, cerebral PaO2 and PaCO2 levels are determined by both native circulation and by ECMO circulation (Figure 1B). In the case of preserved native cardiac function, where the mixing point lies distal to the left subclavian artery, the PaCO2 measured in the right radial artery is likely to represent native blood supply to cerebral circulation, while the abdominal organs are more likely to be supplied by the ECMO circuit (Figure 1C).

Figure 1
Figure 1:
The variable mixing point during different levels of support in peripheral veno-arterial extracorporeal membrane oxygenation. As the mixing point moves, the relative degree to which arch vessels are perfused by native and extracorporeal membrane oxygenation (ECMO) circulations changes.

In clinical practice, the specific position of the mixing point is often unknown, dynamic, and is a function of the competing relative blood flows of the native circulation and the ECMO circuit.

The Impact of Measurement of Blood Gases on the AT

Samples from both right radial artery and postoxygenator are required to reliably determine the cerebral CO2 gas tension, regardless of where the mixing point may be. Thus, during BD testing, the presence or absence of respiratory effort can only be interpreted when:

PaCO2 (right radial) and PbCO2 (postoxygenator) are >60 mmHg (threshold value). This provides conclusive proof of cerebral circulation PaCO2 >60 mmHg.

The Impact of ECMO Configuration

Femoral arterial cannulation is associated with the highest risk of differing arterial gas tensions. Axillary return cannulation has a lower risk of cerebral hypoxia,10 as does central cannulation, due to the proximity of the return cannulas to the cerebral vessels. However, given the uncertain degree and position of the mixing point in all forms of V-A ECMO, measurement of both distal (radial in femoro-femoral, femoral in axillary return) and postoxygenator blood gases will ensure the minimum cerebral CO2 gas tension and O2 saturation can be reliably determined.

Protocol Development and Implementation

The ECMO AT protocol was developed through iterative consultation with specialists from the fields of extracorporeal support systems, organ and tissue donation, intensive care medicine, cardiology, and from other transplant centers in Melbourne. The protocol was instituted at our hospital in 2015 following education for treating clinicians. During the implementation phase clinical determination of BD was paired with radiological testing. The flow diagram for apnea testing is shown in Figure 2. The full protocol is provided in supplemental material (see Appendix 1, Supplemental Digital content,

Figure 2
Figure 2:
Protocol for apnoea testing in patients on ECMO. Apnoea test preparation should include the following: i) ensure all preconditions are met, ii) check baseline PaCO2 from distal arterial gas, iii) ensure ECMO blender and ventilator FiO2 is 1.0, and iv) commence apneic oxygenation. ABG, arterial blood gas; FGF, fresh gas flow; FGF:ECMO, fresh gas flow to extra-corporeal membrane oxygenator blood flow ratio; PaCO2, arterial partial pressure of carbon dioxide; PbCO2, post oxygenator arterial partial pressure of carbon dioxide; SaO2, oxygen saturation.

Three cases where this protocol was used and learnings from each case are described below.


Between 2015 and 2018, there were 45 ECMO-cardiopulmonary resuscitation (E-CPR) episodes in 44 patients. The mortality rate was 32/45 (71%). Majority of E-CPR cases that died were unstable early and therefore not suitable for BD testing. There were three cases where clinical testing was performed. Blood gas results, hemodynamics, and baseline ventilator settings are summarized for the three patients in Table 1 and Figure 3.

Table 1
Table 1:
Results from Clinical Apnea Testing
Figure 3
Figure 3:
Carbon dioxide levels and breathing during the apnoea test.

Each patient was preoxygenated followed by apneic oxygenation via a bag valve mask (BVM) with a positive end expiratory pressure (PEEP) valve adjusted between 5 and 10 cm H2O.

Patient 1 was a male in his sixties, who presented following a prolonged cardiac arrest secondary to a massive pulmonary embolism. He was commenced on peripheral V-A ECMO with a right femoral vein access and left femoral artery return cannula. He failed to show neurologic recovery and on day 3 met all preconditions to perform the AT. After 17 minutes of apnoea, both radial PaCO2 and the postoxygenator PbCO2 were >60 mmHg. This was consistent with BD, although the test was only performed once. The patient underwent a brain perfusion SPECT scan, which confirmed BD.

Patient 2 was a man in his fifties, who was placed on peripheral V-A ECMO via right femoral vein access and left femoral artery return, while receiving CPR for an out-hospital cardiac arrest due to a myocardial infarction. Five days later, BD was suspected, and he met preconditions to perform the AT. After weaning FGF, despite a hyperoxic postoxygenator gas, PaO2 sampled from the right radial artery was low, correlating with a low SaO2, due to native cardiac output and poor lung function. Two rescue breaths were applied via the BVM and adrenaline was decreased. Noradrenaline was escalating to maintain mean arterial pressure >60 mmHg. Despite further reductions in FGF with a concomitant increase in ECMO blood flow in an attempt to overcome native circulation, it was not possible to achieve PaCO2 and PbCO2 >60 mmHg and the patient required significantly more vasopressor support. As a consequence, the AT was abandoned. A nuclear medicine perfusion scan was performed, which was consistent with BD.

Patient 3 was a woman in her forties, who suffered a cardiac arrest secondary to a massive pulmonary embolism. She was placed on peripheral V-A ECMO via right femoral vein access and left femoral artery after 60 mins without spontaneous circulation. On day 3, she met preconditions for BD and clinical testing was performed. While there were no signs of cranial nerve activity, when both PaCO2 and PbCO2 rose above the threshold after approximately 30 minutes, spontaneous respiratory effort was noted consistent with a negative result for clinical BD. Clinical testing was repeated again 2 days later. This time, two ATs performed by independent clinicians confirmed the diagnosis of BD.


We describe the development of a protocol, and its application in 3 patients, to perform an AT for clinical confirmation of BD on V-A ECMO. The protocol highlights the need to measure both radial and postoxygenator blood gases to ensure that the levels of CO2 and O2 in the cerebral circulation are sufficient to interpret the presence or absence of respiratory effort appropriately.

Several previous studies have proposed protocols for performing the AT on ECMO.11,12,14–19 None of these studies have specified an approach to sampling, which would accommodate differences in arterial gas tensions. Hoskote et al.12 suggested it would be possible to turn the ECMO blood flows up to overcome any native circulation. Although theoretically appealing, in our experience this is not possible as there may be access insufficiency before the native circulation can be overcome. Furthermore, increasing the ECMO flow will reduce the FGF:ECMO flow ratio, making gas exchange across the membrane more unpredictable and unreliable.13

Other studies have focused on the issue of preventing hypoxia; however, none of them confirm the PbO2. Studies have suggested turning the FGF to 1 L/min14 or 0.5 L/min,15 while Yang suggested turning off the FGF completely and relying fully on ventilator O2 supply.16 This latter approach will, however, undoubtedly deliver hypoxic blood into the systemic circulation. Giani et al. developed a protocol of apnea testing where they performed recruitment maneuvers and PEEP during the test to prevent hypoxia.17 Hypoxia during the AT can result for several reasons. As the FGF is reduced below 50% of the ECMO blood flow, nonlinear and unpredictable gas transfer across the membrane can result.13 An advantage of our protocol is that it avoids this by adopting a stepwise reduction in FGF followed by regular checking of postoxygenator saturations.

Several other methods to perform the AT on ECMO have been described, including adding CO2 directly to the circuit via a small puncture in the O2 tubing.18,19 While no hypoxia was reported in these studies, the sampling was from arterial samples from unspecified locations and therefore it was not possible to exclude cerebral hypoxia from poorly saturated ECMO blood.

Clinical testing for BD in patients on ECMO is particularly appealing, notwithstanding its complexities. First, hemodynamic instability may make transport of ECMO patients unsafe.13,20 Second, the extracorporeal circuit can reduce the concentration of contrast in the cerebral vessels, leading to reduced quality or even nondiagnostic tests. In addition, the American Academy of Neurology suggests ancillary radiological tests only have a role when clinical testing is not possible.4

Our protocol has several implications. Although both distal arterial PaCO2 and PbCO2 are required to confirm CO2 levels in the brain are >60 mmHg, no additional equipment is required. A “failed” AT (when the patient breaths) avoids unnecessary transport of an unstable patient. Our use of sequential reductions in FGF and postoxygenator blood gas testing avoids unrecognized hypoxia. Finally, this protocol can be applied in all ECMO configurations.

The strengths of this protocol include that it was developed in a high volume ECMO centre with contributions from intensive care, organ donation, and transplantation specialists. The patient examples highlight the robustness of this protocol, with the first two patient examples also undergoing confirmatory imaging and the third patient’s initial test triggering respiratory effort only once both radial and oxygenator CO2 gas tensions had risen above 60 mmHg.

There are several limitations of our protocol. The design of the protocol was from a single centre and constructed on physiologic principles. Application of the protocol requires a distal arterial line (right radial in fem-femoral configuration). Radiological confirmation was not performed in every patient. Lastly, despite the hospital being a high volume ECMO center, BD on ECMO is rare and therefore there were very few patients tested using the protocol; greater uptake is warranted, particularly at other institutions, to further validate this protocol.


Performance of safe and reliable AT while on V-A ECMO requires measurement of both right radial and postoxygenator blood gases to ensure BD is appropriately diagnosed. It is possible to develop and implement a protocol to do this, which is both safe and easy to perform.


1. Lorusso R, Barili F, Mauro MD, et al. In-hospital neurologic complications in adult patients undergoing venoarterial extracorporeal membrane oxygenation: Results from the extracorporeal life support organization registry. Crit Care Med 2016.44: e964–e972.
2. Karagiannidis C, Brodie D, Strassmann S, et al. Extracorporeal membrane oxygenation: evolving epidemiology and mortality. Intensive Care Med 2016.42: 889–896.
3. Magliocca JF, Magee JC, Rowe SA, et al. Extracorporeal support for organ donation after cardiac death effectively expands the donor pool. J Trauma 2005.58: 1095–101; discussion 1101.
4. Wijdicks EF, Varelas PN, Gronseth GS, Greer DM; American Academy of Neurology: Evidence-based guideline update: Determining brain death in adults: Report of the quality standards subcommittee of the American academy of neurology. Neurology 2010.74: 1911–1918.
5. The Anzics Statement on Death and Organ Donation. 2013, 3.2 ed. Australian and New Zealand, Intensive Care Society, p 1–68.
6. Cove ME. Disrupting differential hypoxia in peripheral veno-arterial extracorporeal membrane oxygenation. Crit Care 2015. 19:280, doi: 10.1186/s13054-015-0997-3
7. Hoeper MM, Tudorache I, Kühn C, et al. Extracorporeal membrane oxygenation watershed. Circulation 2014.130: 864–865.
8. Hou X, Yang X, Du Z, et al. Superior vena cava drainage improves upper body oxygenation during veno-arterial extracorporeal membrane oxygenation in sheep. Crit Care 2015.19: 68.
9. Alwardt CM, Patel BM, Lowell A, Dobberpuhl J, Riley JB, DeValeria PA. Regional perfusion during venoarterial extracorporeal membrane oxygenation: A case report and educational modules on the concept of dual circulations. J Extra Corpor Technol 2013.45: 187–194.
10. Wada H, Watari M, Sueda T, et al. Cerebral tissue oxygen saturation during percutaneous cardiopulmonary support in a canine model of respiratory failure. Artif Organs 2000.24: 640–643.
11. Jarrah RJ, Ajizian SJ, Agarwal S, Copus SC, Nakagawa TA. Developing a standard method for apnea testing in the determination of brain death for patients on venoarterial extracorporeal membrane oxygenation: A pediatric case series. Pediatr Crit Care Med 2014.15: e38–e43.
12. Hoskote SS, Fugate JE, Wijdicks EF. Performance of an apnea test for brain death determination in a patient receiving venoarterial extracorporeal membrane oxygenation. J Cardiothorac Vasc Anesth 2014.28: 1027–1029.
13. Lehle K, Philipp A, Hiller KA, et al. Efficiency of gas transfer in venovenous extracorporeal membrane oxygenation: Analysis of 317 cases with four different ECMO systems. Intensive Care Med 2014.40: 1870–1877.
14. Bein T, Muller T, Citerio G. (2019) Determination of brain death under extracorporeal life support. Intensive Care Med 2019. 45: 364–366,
15. Goswami S, Evans A, Das B, et al. (2013) Determination of brain death by apnea test adapted to extracorporeal cardiopulmonary resuscitation. J Cardiothorac Vasc Anesth 2013. 27:312–314, doi: 10.1053/j.jvca.2012.04.020
16. Yang HY, Lin CY, Tsai YT, Lee CY, Tsai CS. Experience of heart transplantation from hemodynamically unstable brain-dead donors with extracorporeal support. Clin Transplant 2012.26: 792–796.
17. Giani M, Scaravilli V, Colombo SM, et al. Apnea test during brain death assessment in mechanically ventilated and ECMO patients. Intensive Care Med 2016.42: 72–81.
18. Champigneulle B, Chhor V, Mantz J, Journois D. Efficiency and safety of apnea test process under extracorporeal membrane oxygenation: The most effective method remains questionable. Intensive Care Med 2016.42: 1098–1099.
19. Pirat A, Kömürcü Ö, Yener G, Arslan G. Apnea testing for diagnosing brain death during extracorporeal membrane oxygenation. J Cardiothorac Vasc Anesth 2014.28: e8–e9.
20. Muralidharan R, Mateen FJ, Shinohara RT, Schears GJ, Wijdicks EF. The challenges with brain death determination in adult patients on extracorporeal membrane oxygenation. Neurocrit Care 2011.14: 423–426.

apnea test; brain death; ECMO

Supplemental Digital Content

Copyright © 2020 by the American Society for Artificial Internal Organs