HYPEROXIA

Jed Wolpaw, MD, M.Ed

Outline
History of Hyperoxia
Benefits of Hyperoxia: SSI?
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Neonates: Not going to cover in detail
Now well established that excess O2 causes harm
 Retinopathy of prematurity
 Bronchopulmonary dysplasia

Gas of choice for neonatal resuscitation is room air

Antioxidant mechanisms in lungs don’t develop until

late gestation
 Preterm infants at high risk for damage

Yee et al, Am J Resp Cell Mol Biol 2013

Question
What were the average PaO2 and %O2 sat amongst 7
physicians breathing ambient air atop Mt. Everest?
A: 24.6 and 54%

B: O and O, they were Hopkins surgery residents and

thus needed no petty comforts like oxygen for survival

C: 39.4 and 68.3%

D: 11.8 and 27.5%

Question
What were the average PaO2 and %O2 sat amongst 7
physicians breathing ambient air atop Mt. Everest?
A: 24.6 and 54%

B: O and O, they were Hopkins surgery residents and

thus needed no petty comforts like oxygen for survival

C: 39.4 and 68.3%

D: 11.8 and 27.5%

How was heart surgery done on infants
prior to bypass?
Hyperbaric chamber
Allowed paO2 in thousands
Could stop heart for a few minutes without tissue
ischemia
What is the Cunningham Ball?
A: An annual dance for very intelligent pigs

B: A hyperbaric chamber the size of a hospital

C: A biennial gala held by the Society for Hyperbaric

Medicine

D: The new term for an underinflated football named

after the New England Patriot’s Ball Boy Greg
Cunningham
What is the Cunningham Ball?
A: An annual dance for very intelligent pigs

B: A hyperbaric chamber the size of a

hospital
C: A biennial gala held by the Society for Hyperbaric
Medicine

D: The new term for an underinflated football named after

the New England Patriot’s Ball Boy Greg Cunningham
The Cunningham Ball
 Surgical Site Infection?
JAMA 2004: RCT 165 pts undergoing abdominal surgery
Randomized to 80% or 35% Oxygen during surgery and
first two hours after
100% allowed during induction and emergence or if
SaO2 fell below 94%
Increase in SSI in group receiving 80% (25% vs 11.3%
p=0.02)
Criticism: Groups differed in intraop blood loss, post-op
fluid replacement, rates of obesity, study stopped early

Pryor, et al, JAMA 2004

 Surgical Site Infection Meta-analysis

Excluding largest study

(1000 pts per arm, more
than other 4 combined)

Qadan et al, Archives of Surgery 2009

 Some Thoughts
Evolved adaptations: Hypoxia vs. hyperoxia
How low is too low?

Average PaO2 24.6 (19.1-29.5); SaO2 54%

(34.4-69.7); PCO2 13.3 (10.3-15.7)
Grocott, et al, NEJM 2009
 Current Practice
Post-MI, post-arrest, post-stroke pts typically get
supplemental O2 regardless of SaO2 or PaO2
Suzuki, J Crit Care 2013
 51 patients over 358 ICU mechanically ventilated days
 Most patients (59%) had SaO2>98% w PaO2 80-120
 26% patients had PaO2>120 and 4% >202
 Only 1% of patients had FiO2 0.21-0.29
 Clinicians were more concerned with FiO2 than PaO2
Almost no adjustments were made regardless of how high the
PaO2 was if the FiO2 was 0.3-0.5

Damiani et al, Crit Care 2014; Suzuki, et al, J Crit Care 2013
 Data Caveat
Almost all of the data are from Observational studies
Lack of RCTs comparing two different SaO2 or PaO2
levels
Different studies use different definitions of hyperoxia
and measure it at different times

Damiani et al, CCM 2014

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Reactive Oxygen Species
Examples of ROS
 Superoxide
 Hydrogen Peroxide
 Peroxynitrite
 Hydroxyl radical
Some serve important roles but in excess can:
 Interact with and damage DNA, lipids, proteins,
carbohydrates
Concentration of ROS in exhaled gas increases after 1
hour of breathing 28% O2

Brown and Griendling, Circ Res 2015

 Reactive Oxygen Species

Brown and Griendling, Circ Res 2015

 Hyperoxic Hypocapnia
ROS are thought to account for some of increased
ventilation seen with increasing levels of O2

ROS may act at respiratory centers in brainstem

Causes reduction in PaCO2

Brown and Griendling, Circ Res 2015

 ROS Benefits
Known to play role in defense against pathogens and phagocytosis
Emerging evidence that ROS play an important role in other pathways

Isoproterenol induced increases

in calcium flux and contractility
were attenuated by NAC
B-adrenergic activation caused
increase in ROS
Prolonged activation (>24h)
caused cell death
Excess ROS production caused
opposite results
ROS in limited amounts play
important role

Andersson et al, J Phys 2011

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Two hit model
Ball and Ranzani, Intensive Care Medicine Jan 2015
First hit: Hypoxic ischemic injury from cardiac arrest
Second hit: Reperfusion injury made worse by:
 Hyperoxic vasoconstriction (Hypocapnia and ROS
scavenging NO)
 Increased susceptibility to ROS (increased ROS from
hyperoxia)

Ball and Ranzani, Int Care Med 2015

 Hyperoxic Hypocapnia
Mechanism not fully understood
Initial brief hypoventilation followed by
hyperventilation
Possible mechanisms
 O2 induced vasoconstriction in brain leading to reduced
blood flow and subsequent hyperventilation
 Reverse Haldane effect: Increased O2 causes reduced
binding of CO2 to Hgb in tissues (and respiratory
center) leading to locally increased PaCO2 and
hyperventilation

Sjoberg et al, J Int Med 2013

 Restrospective Review
184 pts over 2 year period (2008-2010) with cardiac
arrest who survived at least 24h (60% PEA, 40%
VT/VF; 60% out of hospital)
36% exposed to severe hyperoxia (PaO2>300)
Adjusted OR for reduced survival per hour of exposure
in this group was 0.83 (0.69-0.99) p=0.04
No difference in neurologic outcomes
Several limitations: retrospective, no data on
important comorbidities, LOS, epi dose

Elmer et al, Int Care Med 2015

 Elmer Review graphs

Note, this is after

ROSC. They all got
100% O2 during
CPR

Elmer et al, Int Care Med 2015

 RCT 30% vs. 100% for post-ROSC
Pilot study, not powered for outcomes
14 pts in each arm
30% vs. 100% O2 for 60 minutes post-ROSC
Measured serum markers of neuronal injury (NSE and
S-100), adequacy of oxygenation and need to increase
FiO2 in 30% group
Found trend toward increased serum markers in 100%
group (statistically significant when looking only at pts
not getting hypothermia)
No hypoxemia in 30% group

Kuisma et al, Resuscitation 2006

 Kilgannon JAMA 2010
Database of 120 US ICUs from 2001-2005
Non-traumatic cardiac arrest w CPR within 24h of ICU
arrival
3 groups based on first ABG w hypoxic PaO2<60 (or p/f
ratio <300), normoxic 60-299 and hyperoxic >300
6326 pts, 18% hyperoxic, 63% hypoxic
Mortality hyperoxic 63%, hypoxic 57%, normoxic 45%
Limitations: Observational only (association, not
causation), No data on actual resuscitation (duration,
etc.), no data on who got hypothermic treatment
Take away: Hypoxia and hyperoxia are probably both bad
Kilgannon et al, JAMA 2010
 Kilgannon survival curve

Kilgannon et al, JAMA 2010

 Damiani Meta Analysis Post-cardiac arrest
Hyperoxia defined as PaO2>300

Damiani et al, Crit Care 2014

 Wang Meta Analysis
14 Studies, large heterogeneity
OR for death was 1.4 (1.02-1.93) I^2=69%. Hyperoxia
defined as PaO2>300

Wang et al, Resuscitation 2014

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Cardiac Ischemia and O2 history
1950 Russek et al reported supplemental o2 failed to
reduce ECG signs of ischemia or reduce angina
1969 Bourassa et al proposed that the decreased blood
flow from hyperoxia may cause ischemia
1976 RCT Rawls and Kenmure showed increased
cardiac markers in acute MI pts getting high flow o2 vs
room air
 Also trend toward tripled death rate.

Iscoe et al, Critical Care 2011

 Downside of O2 for ischemic hearts
Recommendations and Guidelines have traditionally
called for O2 to be administered regardless of PaO2 or
SaO2 (MONA)
However, emerging evidence suggests it may:
 Reduce coronary blood flow
 Reduce stroke volume and cardiac output
 Increase SVR
 Increase reperfusion injury

Sjoberg et al, J Int Med 2013

 Reduced O2 delivery
Increasing PaO2 from 100 to 600 increases content by
only 15ml/L or about 7-8%
Hyperoxia decreases cerebral blood flow by 11-33% and
coronary flow by 8-29%
Net decrease in delivery may be cause of worse
outcomes
Theory that hyperbaric O2 may be able to overcome
this but data is lacking

Iscoe et al, Critical Care 2011

 Cochrane Review 2010
Looked for RCTs of NSTEMI or STEMI comparing air
with O2
Found 4 trials, 430 pts, 17 deaths, pooled RR for death
was 2.11 (0.78-5.68)

Cabello et al, Cochrane Collaboration 2010

 Cochrane Review 2010

Cabello et al, Cochrane Collaboration 2010

 Ongoing Trial
AVOID study, Australia, Multicenter RCT, enrolling
490 pts w STEMI without hypoxia

Randomized to room air vs supplemental O2

Primary outcome is infarct size

Stub et al, Am Heart Journal 2012

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Endothelial dysfunction
Hyperoxia is efficacious in non-clinical studies but this
has not translated clinically
Knockout mice designed to mimic endothelial
dysfunction found in acute stroke
Normal mice: hyperoxia improved tissue perfusion
Knockouts: Hyperoxia worsened perfusion
One knockout model was nitric oxide synthase
 Suggests that inadequate NO may place pts at increased
vulnerability to ROS from hyperoxia

Shin et al, J Neuroscience 2014

 Post-stroke mortality
Restrospective multicenter cohort study 2894 pts
Acute ischemic stroke, SAH and ICH in 84 ICUs
Hyperoxia (PaO2>300), Hypoxia (PaO2<60) and
Normoxia
Crude OR for hyperoxia 1.7 (1.3-2.1 p<0.0001) and for
hypoxia 1.3 (1.1-1.7 p=0.01)
Adjusted OR for hyperoxia 1.2 (1.04-1.5)
Limitations: Retrospective, observational, lacking data
on disease severity

Rincon et al, Crit Care Med 2014

 Post-stroke mortality

Rincon et al, Crit Care Med 2014

 Stroke Meta Analysis
Hyperoxia defined as PaO2>300

Damiani et al, Crit Care 2014

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Hyperoxia in TBI
Retrospective multicenter cohort study of 1212 pts in 61
ICUs
Divided into hyperoxia (PaO2>300), hypoxia (PaO2<60)
and Normoxia
21% were hyperoxic (of these half had PaO2>400), 46%
were hypoxic
Adjusted OR for death in hyperoxia group was 1.5 (1.02-
2.4) p=0.04
Proposed mechanism is 11-33% decrease in cerebral
blood flow with hyperoxia, maybe worse with endothelial
dysfunction
Rincon et al, J Neurol, Neurosurg, Psychiatry 2014
 Hyperoxia in TBI

Rincon et al, J Neurol, Neurosurg, Psychiatry 2014

 Excitotoxicity in TBI
Retrospective analysis of 1130 TBI patients monitored
with cerebral microdialysis (CMD) and and brain tissue
oxygenation (PbtO2)
Increased FiO2 is common way to treat decreased PbtO2
in TBI but recent data suggests it may cause harm
Increased exctracellular glutamate increases
excitotoxicity of brain tissue and cell death
In culture astrocytes exposed to high FiO2 had reduced
ability to protect other cells from glutamate
CMD allows measurement of extracellular glutamate

Quintard, Neuro Crit Care 2014

 Excitotoxicity in TBI

Levels of glutamate greater

than 10 umol/l have been
shown to correlate with
worse neurologic outcome
after TBI

Quintard, Neuro Crit Care 2014

 TBI Meta Analysis

Damiani et al, Crit Care 2014

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Pulmonary toxicity history
1775 Joseph Priestly: “Though pure dephlogisticated air
might be very useful as a medicine, it might not be
proper for us in the usual healthy state to the body.”
1899 Lorrain-Smith exposed animals to 3 atm O2 and
found their lungs to be severely damaged
Mid 1900’s new more efficient ventilators led to
increasing pulmonary oxygen toxicity in humans

Ooij et al, Res Phys and Neurobiology 2013

 Symptoms of oxygen toxicity
Mild substernal tickling, tracheal irritation

Chest tightness leading to increasing substernal pain

and uncontrolled coughing

DOE and eventually dyspnea at rest

Ooij et al, Res Phys and Neurobiology 2013

 Pathophysiologic changes
Denuded alveolar type 1 cells
Capillary cell edema
Squamus metaplasia of tracheal and bronchial mucosa
Eosinophilic slough within bronchioles
Formation of hyaline membranes
End stage of emphysematous destruction and/or
fibrosis

Ooij et al, Res Phys and Neurobiology 2013

 Clinical changes
Impaired mucociliary function leading to mucous
plugging, atelectasis, secondary infection
Erythema and edema of large airways can be seen
bronchoscopically after 6 hours at 90% O2
Concentration of ROS in exhaled gas can be seen after
1 hour breathing 28% O2
Absorptive atelectasis can cause 11% shunt in healthy
people breathing 100% O2 for 30 minutes

Ooij et al, Res Phys and Neurobiology 2013

 Cellular signaling pathways

Porzianato et al, Resp Phys and Neurobiology 2014

 Murine Knockouts
Superoxide dismutase knockout mice die shortly after
birth with extensive ROS injury in lungs, heart, brain
CYP1A 1 and 2 knockouts show severe damage in
response to hyperoxia

Jiang et al, Biochem Biophys 2012

 Pathophysiologic changes

Ooij et al, Res Phys and Neurobiology 2013

 Measuring pulmonary toxicity

Unit of Pulmonary Toxicity Dose: t is duration of

exposure and PO2 is inhaled concentration of O2
 Used by Divers worldwide to calculate exposure
 Correlates to decrease in FVC

Ooij et al, Res Phys and Neurobiology 2013

 Spirometry
FVC decreases
 Likely first from tracheobronchitis causing pain with
inspiration
 Later due to pulmonary edema, thickened airway
epithelium
FEV1, DLCO decrease, DLCO for days to weeks after
exposure in some people
Pulmonary compliance decreases
Early exudative phase: changes are reversible
Late proliferative phase: changes can be permanent

Ooij et al, Res Phys and Neurobiology 2013

 Special cases
Patients with pre-existing injury from Bleomycin,
Amiodarone and external beam radiation
 At higher risk of developing diffuse alveolar damage
 Likely secondary to a preexisting “First hit” as Bleomycin
damages DNA and prevents lung’s ability to heal from
oxidative damage
Neonatal exposure to hyperoxia places mice at
increased risk of severe bleomycin and influenza
damage to lungs

Yee et al, Am J Resp Cell Mol Biol 2013

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 COPD high flow vs titrated O2
405 Patients in Australia who called EMS for COPD
exacerbation
Randomized to high flow O2 (10L NRB face mask) vs.
titrated O2 to keep sats 88-92%
For all patients RR of death was 0.42 w titrated O2
(0.2-0.89 p=0.02)
For confirmed COPD RR was 0.22 (0.05-0.91 p=0.04)
Titrated O2 group less likely to have respiratory
acidosis or hypercapnia
NNH 14 (for every 14 pts given high flow, one will die)

Austin et al, BMJ 2010

 COPD high flow vs titrated O2
Found many providers gave higher flow O2 than
dictated because of “ingrained habit”
 Effect may have been larger if less of this crossover
Guidelines in Britain and U.S.A. now suggest titrated
O2 rather than high flow in COPD

Austin et al, BMJ 2010

 Hyperoxia and acidosis in COPD
In a study of 980 pts in
ED w COPD exacerbation
 20% had respiratory
acidosis
 Strong correlation with
PaO2
 Presence of respiratory
acidosis was associated
with 12.1% risk of death
compared to 6.9%
without

Lellouche et al, Can Resp Journ 2013

 Closed loop O2 delivery for COPD
Uses SaO2 to provide continuous feedback and allows
continuous adjustment of FiO2 to maintain a
programmed SaO2
3 systems have been developed, being tested currently

Lellouche et al, Can Resp Journ 2013

 Closed loop O2 delivery for COPD

Lellouche et al, Can Resp Journ 2013

Outline
History of Hyperoxia
Benefits of Hyperoxia
Risks of Hyperoxia
 ROS
 Post-Cardiac Arrest
 Post-MI
 Stroke
 TBI
 Lung Injury
 COPD
 Intraoperative
Summary
 Time to desaturation

Lumb and Walton, Anesthesiology 2012

 One Lung Ventilation
Lung collapses faster with 100% O2 than with air
Still faster with N2O
Most useful with bronchial blockers
High risk for lung injury with stress on single lung and
typically high FiO2
Recommendations are to start with 80% and titrate
down to maintain SaO2 90-92%
Non-ventilated long gets hypoxic and is very
vulnerable to reperfusion injury exacerbated by high
FiO2

Lumb and Walton, Anesthesiology 2012

 PONV
Early studies showed decreased incidence of PONV
with 80% vs. 30% FiO2
Subsequent studies have failed to replicate this
Meta Analysis 2008 A&A showed no benefit

Lumb and Walton, Anesthesiology 2012

 Anesthetic outcomes in children
Kopp suggests that hyperoxia during anesthetics may
play a role in the potential harm caused by repeated
anesthetics in children under age 2
Suggests future studies should include a 21% FiO2 arm

Kopp, Anesthesiology 2009

Overall Intraop
Risk for lung injury, atelectasis, decreasing lung
volumes, airway irritation and postoperative pain,
local tissue hypoxia, coronary vasoconstriction,
cerebral vasoconstriction, etc.
No good data for reduced n/v, infection (except maybe
in colon surgery)
Probably worthwhile for induction and possibly
emergence
Otherwise, balance risk of losing the airway with the
risks of high FiO2
Summary
Oxygen can be life-saving
It can also be harmful
 Excess ROS
 Vasoconstriction causing reduced local DO2
 Hypocapnia
 Direct tissue damage
We are not starting with an even playing field
Try not to reflexively give high flow O2 in traditional
situations (Post-arrest, MI, stroke, TBI, COPD)
With some exceptions, titrate O2 to normoxia
More and better studies are needed