Resuscitation 83 (2012) 403 404
Contents lists available at SciVerse ScienceDirect
Resuscitation
jo u rn al hom epage : www.elsevier.com/locate/resuscitation
Editorial
Hyperoxia post-cardiac arrest  The debate continues?
Concerns about the use of high concentrations of inspired oxy- baseline characteristics were comparable, a marked difference in
ć% ć%
gen have provoked considerable debate. Detrimental effects of lowest median body temperature was noted (ANZ 34.9 C, US 36 C).
hyperoxia are well established in neonatal resuscitation.1 Exper- This presumably represents a far higher uptake of therapeutic
imental and conflicting observational data have fuelled discussion hypothermia in Australasia and is more representative of current
in adult post-cardiac arrest practice.2,3 The safety of exposing non- clinical practice. It could also explain the 50% increase in favourable
hypoxaemic patients to high fraction of inspired oxygen (FiO2) in outcome seen in the ANZ study (65% discharged home, 44% in the
the context of acute myocardial ischaemia and stroke4 has also US). It may however, provide an explanation for the difference in
been questioned and appropriate arterial blood partial pressure of primary outcome. Hypothermia is known to mitigate reperfusion
oxygen (PaO2) targets are being reconsidered in numerous settings injury and could feasibly reduce the magnitude of effect of hyper-
including traumatic brain injury, acute lung injury and even the oxaemia on mortality. This hypothesis was also suggested by a
general critical care patient population.5 7 small randomised controlled pilot trial of 28 patients resuscitated
In post cardiac arrest care the debate has been dominated by from out-of-hospital cardiac arrest.11 Patients were ventilated with
the publication of two large retrospective database analyses, one either 30% or 100% O2 for 60 min post return of spontaneous circu-
using the Project IMPACT database in the United States (US) and lation (ROSC). The use of 100% O2 was associated with an increase
the second the Australian and New Zealand Adult Patient Database in neuron specific enolase (NSE) at 24 h in a subgroup of patients
(ANZ-APD).8,9 After a multivariate analysis, the US investigators not treated with therapeutic hypothermia. As NSE is an estab-
found that post-resuscitation hyperoxaemia was an independent lished marker of neuronal injury, this result also suggests that
predictor of in-hospital mortality (odds ratio 1.8; 95% CI 1.2 2.2) hypothermic conditions may protect against hyperoxygenation-
compared with normoxaemia or hypoxaemia. In a secondary anal- generated reperfusion injury. This trial was underpowered to make
ysis of this data the authors demonstrated a dose-dependent any assessment of outcome or survival.
association of hyperoxaemia with mortality and independent sta- In the context of this controversy, it is useful to revisit the
tus at hospital discharge. For every 100 mmHg rise in PaO2, experimental data on which early concern was based. In this
mortality increased by 24% (OR 1.24; 95% CI 1.18 1.24).10 issue of Resuscitation, Pilcher et al. present a timely summary of
The ANZ study considered a larger, more complete data set the available animal studies and their clinical applicability.12 In
(5.4% excluded versus 27% in the US study) generated from 12,108 a meta-analysis of six included studies (n = 95) they report that
patients across 125 intensive care units (ICUs). It was rigor- resuscitation with 100% oxygen resulted in significantly worse neu-
ously conducted with multifaceted assessments, which included rological deficit score (NDS) than oxygen administered at lower
adjustment for illness severity. Hyperoxaemia was found to be concentrations in a variety of animals following experimental car-
relatively uncommon. No robust or consistently reproducible diac arrest, by a factor of approximately two-thirds of a standard
relationship with mortality could be demonstrated. The authors deviation (standardised mean difference of -0.64; 95% CI -1.06 to
cautioned against polices of deliberately reducing FiO2 because of -0.22). Interpreting this quantitative meta-analysis must be done
the risks of precipitating the well-established adverse effects of with caution given the heterogeneity of the included studies, the
hypoxaemia. different species studied, the differences in FiO2 strategies, the
Delivery of higher FiO2 occurs in sicker patients and hyper- small numbers and methodological weaknesses in the remaining
oxaemia could therefore simply constitute some other marker of publications. The authors acknowledge the questionable inclusion
illness severity; however, there are several important differences of the study by Yeh et al., which was primarily designed to consider
between these two studies that are also worth highlighting. Firstly, the effects of oxygenation strategy during CPR, and administered
the US study used the first blood gas at an unspecified time within 100% oxygen to all animals after ROSC.13 However, omitting this
24 h of admission whereas the authors of the ANZ study chose the study in a sensitivity analysis did not significantly affect the results.
worst blood gas in the first 24 h. When considering the effect of The Lipinski et al. study also used a very different asphyxial model
hyperoxaemia, neither could be considered ideal. Although the ANZ of cardiac arrest, which may have impacted significantly on the
study attempted to correlate the worst blood gas with the mean findings.14 This is of particular relevance given that this is the only
PaO2 in the first 24 48 h, it remains likely that a significant propor- animal model to date that has failed to demonstrate an adverse
tion of patients exposed to hyperoxaemia were not identified in affect of hyperoxic resuscitation.
either study. Moreover, duration and timing of exposure, which Despite these reservations a magnitude of effect is clearly
may be critical, cannot be considered. Secondly, although most demonstrated and cannot be ignored. The results were correlated
0300-9572/$  see front matter © 2012 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.resuscitation.2012.01.035
404 Editorial / Resuscitation 83 (2012) 403 404
in a number of the studies with increased histological neuronal cell 1. Conflicts of interest statement
damage and evidence of impaired cerebral metabolic function, in
addition to reversal by antioxidant pretreatment.12 These results None.
are supported by additional animal studies excluded from this
review because they considered a global cerebral ischaemia model References
without inducing full cardiac arrest. These studies also consistently
1. Davis PG, Tan A, O Donnell CPF, Schulze A. Resuscitation of newborn infants
demonstrated that early exposure to 100% oxygen increases oxida-
with 100% oxygen or air: a systematic review and meta-analysis. Lancet
tive stress, worsens neuronal cell death and has a detrimental effect
2004;364:1329 33.
on short-term neurological function.15 17
2. Neumar RW. Optimal oxygenation during and after cardiopulmonary resuscita-
tion. Curr Opin Crit Care 2011;17:236 40.
How do we interpret these findings? Whilst the consistency
3. Hoedemaekers CW, van der Hoeven JG. Hyperoxia after cardiac arrest may not
of results across a number of species is compelling, the animal
increase ischemia-reperfusion injury. Crit Care 2011;15:166.
models do not accurately replicate the post-cardiac arrest clini-
4. Iscoe S, Beasley R, Fisher JA. Supplementary oxygen for nonhypoxemic patients:
O2 much of a good thing? Crit Care 2011;15:305.
cal situation. Timing of exposure to hyperoxaemia varies, often
5. Diringer MN. Hyperoxia: good or bad for the injured brain? Curr Opin Crit Care
from the onset of CPR and in some cases prior to cardiac arrest,
2008;14:167 71.
with a duration of exposure from 10 to 60 min post reperfusion.
6. Altemeier WA, Sinclair SE. Hyperoxia in the intensive care unit: why more is not
always better. Curr Opin Crit Care 2007;13:73 8.
None of the studies consider the effect of therapeutic hypothermia,
7. de Jonge E, Peelen L, Keijzers PJ, et al. Association between administered oxy-
increasingly a standard of care in many ICUs. This may influence
gen, arterial partial oxygen pressure and mortality in mechanically ventilated
significantly the effects of hyperoxaemia. Crude NDS in animals is
intensive care unit patients. Crit Care 2008;12:R156.
not a clinically tangible measure when considering advanced cog- 8. Kilgannon JH, Jones AE, Shapiro NI, et al. Association between arterial hyper-
oxia following resuscitation from cardiac arrest and in-hospital mortality. JAMA
nitive function in humans and effect on NDS was only assessed
2010;303:2165 71.
1 4 days post ROSC. The animal studies to date have failed to con-
9. Bellomo R, Bailey M, Eastwood GM, et al. Arterial hyperoxia and in-hospital
sider or demonstrate a significant effect on longer-term outcome. mortality after resuscitation from cardiac arrest. Crit Care 2011;15:R90.
10. Kilgannon JH, Jones AE, Parrillo JE, et al. Relationship between supranormal oxy-
It must also be emphasized that the experimental data reflect the
gen tension and outcome after resuscitation from cardiac arrest. Circulation
effect of hyperoxaemic reperfusion on ischaemic brain injury in
2011;123:2717 22.
otherwise healthy animals. Clearly, this is very different to the
11. Kuisma M, Boyd J, Voipio V, Alaspää A, Roine RO, Rosenberg P. Comparison of
30 and the 100% inspired oxygen concentrations during early post-resuscitation
clinical situation involving typically elderly patients with multi-
period: a randomised controlled pilot study. Resuscitation 2006;69:199 206.
ple co-morbidities. Post-cardiac arrest syndrome constitutes not
12. Pilcher J, Weatherall M, Shirtcliffe P, Bellomo R, Young P, Beasley R. The effect
only brain injury but myocardial dysfunction, a systemic ischaemia
of hyperoxia following cardiac arrest  a systematic review and meta-analysis
reperfusion (IR) response and persistent precipitating pathology.18 of animal trials. Resuscitation 2012;83:417 22.
13. Yeh ST, Cawley RJ, Aune SE, Angelos MG. Oxygen requirement during car-
Hyperoxaemia may well have effects on each of these processes and
diopulmonary resuscitation (CPR) to effect return of spontaneous circulation.
global end-organ function. As cardiac arrest in adults often occurs
Resuscitation 2009;80:951 5.
14. Lipinski CA, Hicks SD, Callaway CW. Normoxic ventilation during resuscitation
in the context of an acute coronary event, the myocardial effects of
and outcome from asphyxial cardiac arrest in rats. Resuscitation 1999;42:221 9.
hyperoxia, including coronary vasoconstriction, reduced coronary
15. Mickel HS, Vaishnav YN, Kempski O, von Lubitz D, Weiss JF, Feuerstein G. Breath-
blood flow and impaired myocardial mitochondrial function, may
ing 100% oxygen after global brain ischemia in Mongolian Gerbils results in
all have significance.19,20 increased lipid peroxidation and increased mortality. Stroke 1987;18:426 30.
16. Feng ZC, Sick TJ, Rosenthal M. Oxygen sensitivity of mitochondrial redox status
How do we progress from here? Oxygen has a complex
and evoked potential recovery early during reperfusion in post-ischemic rat
biological role. We are only just beginning to understand the
brain. Resuscitation 1998;37:33 41.
mechanisms that have evolved to enable the body to adapt and 17. Hazelton JL, Balan I, Elmer GI, et al. Hyperoxic reperfusion after global cerebral
ischemia promotes inflammation and long-term hippocampal neuronal death.
survive hypoxic insults and the variation in individual, organ-
J Neurotrauma 2010;27:753 62.
based and disease-related responses to both low and supernormal
18. Nolan JP, Neumar RW, Adrie C, et al. Post-cardiac arrest syndrome: epidemiol-
PO2.21,22 Without a simple means of monitoring the adequacy
ogy, pathophysiology, treatment, and prognostication. A Scientific Statement
from the International Liaison Committee on Resuscitation; the American
of regional tissue perfusion and oxygenation it is difficult to
Heart Association Emergency Cardiovascular Care Committee; the Council on
define and prevent harm from both hypoxaemia and hyperox-
Cardiovascular Surgery and Anesthesia; the Council on Cardiopulmonary, Peri-
aemia.
operative, and Critical Care; the Council on Clinical Cardiology; the Council on
Stroke. Resuscitation 2008;79:350 79.
There is clearly a limit to how much further useful information
19. Farquhar H, Weatherall M, Wijesinghe M, et al. Systematic review of studies of
can be provided by animal studies of cardiac arrest. Insight may
the effect of hyperoxia on coronary blood flow. Am Heart J 2009;158:371 7.
still be gained by determining the impact of hyperoxaemia beyond
20. Angelos MG, Yeh ST, Aune SE. Post-cardiac arrest hyperoxia and mitochondrial
function. Resuscitation 2011;82(Suppl. 2):S48 51.
the first hour post ROSC, examining effects on long-term survival
21. Eltzschig HK, Carmeliet P. Hypoxia and inflammation. N Engl J Med
and considering how therapeutic hypothermia may impact on the
2011;364:656 65.
effects of hyperoxia. It seems likely, given the cumulative inter-
22. Martin DS, Khosravi M, Grocott MP, Mythen MG. Concepts in hypoxia reborn.
est in the effects of oxygen exposure in various patient groups, Crit Care 2010;14:315.
that we will gain valuable clinical insight from additional studies
in the setting of acute myocardial infarction, stroke and general Clare Hommers
ICU patients. The available experimental and clinical data certainly Anaesthesia and Intensive Care Medicine, Bristol
gives cause for concern. However, only well-designed randomised School of Anaesthesia, Severn Deanery,
controlled clinical trials will aid understanding of any association United Kingdom
between hyperoxaemia and mortality and enable us to evaluate E-mail address: [e-mail]
the safety and efficacy of a controlled reoxygenation strategy post-
cardiac arrest. In the mean time clinicians must, as ever, weigh up 29 January 2012
the available evidence and consider the relative risks and benefits
in individual patients before turning down the FiO2.