Damaged brains don't show an oxygen preference (above "enough")
Ill LOGICAL patients kind of maybe almost prove it
Post-cardiac arrest research and care delivery are guided by the hope that a therapeutic window exists between an episode of hypoxic brain injury and irreversible neuronal death, i.e., permanent brain damage.
This is entirely rational, because hypoxic-ischemic brain injury is a complex process that proceeds in phases: the initial hypoxic insult (occurring over minutes between arrest and ROSC), followed by secondary damage from the inflammatory processes of reperfusion injury.
In ischemic stroke, for example, an at-risk but viable penumbra of salvageable brain tissue can be visualized surrounding the ischemic core of an infarcted vessel’s territory. Thrombectomy can restore perfusion, saving these at-risk areas from tissue death.
No analogous zone of revivable brain can be visualized from the more diffuse injuries resulting in hypoxic-ischemic encephalopathy. Still, animal models have suggested that interventions (e.g., induced hypothermia) might mitigate damage from the secondary post-reperfusion processes that evolve over days after cardiac arrest.
And at least one intervention that might worsen that damage: excessive oxygen delivery has been suspected of producing toxicity via reactive oxygen species generation that overwhelms antioxidant defenses, amplifying reperfusion injury and neuronal death.
Hyperoxia after cardiac arrest has been associated with worse neurologic injury and increased neuronal death in experimental animals.
In humans, the subset of patients in the PILOT trial with cardiac arrest who received conservative oxygen (target <96% SpO2) had almost double the rate of a good neurologic outcome compared to those targeted to SpO2 96%-100%.
Patients in the ICU-ROX trial with cardiac arrest had greater 6-month survival and a trend toward improved neurologic outcomes.
On the other hand, hypoxemia carries obvious risks of worsening an evolving brain injury. Targeting lower oxygen targets might allow a portion of patients to slip into harmful levels of hypoxemia.
The LOGICAL trial was conducted to discern the potential benefits or harms of lower oxygenation targets in patients suffering cardiac arrest.
At 53 ICUs in Australia, New Zealand, and Ireland, 1,840 patients who were comatose and vented after cardiac arrest were randomized to either conservative (target SpO2 90-95%, allowed to receive room air) or liberal oxygen therapy (target SpO2 90-100%, minimum FiO2 0.3).
This was elegantly nested within the Mega-ROX research program (an ongoing megatrial/platform testing oxygenation strategies in critical illness).
About three quarters had out-of-hospital cardiac arrest, roughly half had a shockable first rhythm, and the median time from ROSC to randomization was about 7 hours.
LOGICAL Results
The two groups achieved adequate separation, with the conservative group spending substantially less time with SpO₂ ≥97% and much more time breathing room air.
More liberal oxygen patients had hyperoxia (PaO2 >100mm Hg) (78% vs. 58%), and more conservative patients had episodes of hypoxemia (PaO2 <60 mm Hg), 43% vs 28%. Time spent with SpO2 < 88% was a median of zero hours in both groups.
But at 180 days, the rate of survival with a favorable functional outcome (Extended Glasgow Outcome Scale score of 5 to 8) was nearly identical at 38.2% in the conservative group vs. 39.7% in the liberal arm.
If anything, the point estimate slightly favored liberal oxygen, though the confidence interval was compatible with modest benefit or modest harm.
Secondary outcomes were also similar: survival at 180 days was 48.0% with conservative oxygen and 49.7% with liberal oxygen. Quality-of-life scores, cognitive outcomes, mechanical ventilation duration, ICU length of stay, hospital length of stay, and discharge home were all similar.
Subgroup analyses did not identify a group clearly benefiting from conservative oxygen. Results were consistent by arrest location, initial rhythm, medical versus nonmedical cause, and earlier versus later randomization.
Conclusions
Brain injury evolves for hours or days after the sudden diffuse ischemic injury caused by hypoperfusion during cardiac arrest.
However, this secondary process of reperfusion injury has thus far been stubbornly resistant to intervention. Induced hypothermia was strongly believed to mitigate reperfusion injury, until multiple randomized trials failed to show a benefit.
Based on the LOGICAL trial, it is exceedingly unlikely that avoidance of hyperoxia can make a meaningful difference, either.
That being said, since most patients were randomized more than six hours after achieving ROSC, the LOGICAL trial is vulnerable to the same criticisms of most of the negative hypothermia trials: the intervention may have been delivered too late to make a meaningful difference, and this isn’t proof that conservative oxygen doesn’t help.
In the larger context of an absent signal for benefit from differential oxygenation strategies, however, most intensivists will likely log the trial results as supporting a logical heuristic of managing oxygenation for vented cardiac arrest patients: not too high, not too low, with the knowledge that almost all of the eventual outcome has already been determined before the patient’s arrival to the ICU.