Hyperoxia and acute brain injury

Dear editor,

We read the interesting work of Romero‑Garcia and colleagues on the association between partial pressure of O₂ in arterial blood (P_aO₂) and neurological outcomes in a context of acute brain injury (ABI) [1]. This is a particularly relevant topic as the brain is highly vulnerable to excessive oxidative stress and reactive oxygen species (ROS) due to its huge dioxygen (O₂) consumption and its high polyunsaturated fatty acid and iron content. The authors showed that high P_aO₂ (hyperoxemia) was associated with poor neurological outcomes and mortality and conclude that it is important to adjust oxygenation strategies in this population [1]. It gives us the opportunity to explore the physiological effect of O₂, hyperoxemia and hyperoxia on cerebral blood flow and metabolism beyond the sole results of the present meta-analysis.

Hyperoxemia, hyperoxia and reactive oxygen species

Oxidative stress via the generation of reactive oxygen species (ROS) is one of the suggested mechanisms to explain the worsened outcomes associated with hyperoxemia [1, 2]. To properly understand how O₂ affects the generation of ROS, it seems important to recall some fundamental of physiology. ROS are a normal product of aerobic metabolism and are generated inside the cells, mainly in the mitochondria [2, 3]. Their excessive production is observed in various conditions, but mostly when the O₂ concentration reaching the cells abruptly changes as it happens during ischemia-reperfusion phenomenon (i.e., generation of superoxyde anion by the degradation of the accumulated hypoxanthine during ischemic phase by the xanthine oxidase following reestablishment of O₂ supply) [4, 5].

However, to hypothesize that hyperoxemia leads to increase ROS is an oversimplified statement [6]. Hyperoxemia refers by definition to a supranormal quantity of O₂ in the blood (assuming that a normal quantity is observed in a healthy person breathing air at sea level). It usually refers to either the partial pressure of dissolved O₂ (the P_aO₂ in clinical practice) or the arterial content in O₂ (pooling both the dissolved O₂ and the O₂ carried by hemoglobin). Hyperoxia refers to a supranormal concentration of O₂ reaching the intracellular compartment. In normobaric (and normothermic) conditions, the absolute quantity of O₂ transported in its dissolved form is very low due to the low solubility coefficient of O₂ in blood (Fig. 1). This is why a carrier system (the hemoglobin) is needed to supply the global demand in O₂ [7]. However, only dissolved O₂ molecules can diffuse through the tissue to reach their effect site [7].

Fig. 1.

O₂ journey: from the airways to the mitochondria. Figure 1 represents a schematic representation of the concentration of free O₂ molecules during their journey from the atmosphere to the cells in a healthy subject. Airway’s cells are directly exposed to O₂ molecules whose partial pressure varies linearly with the fraction of inspired O₂. Then O₂ molecules diffuse into the capillaries blood where they are loaded by hemoglobin. In standard condition of pressure and temperature, the solubility of O₂ in blood is very low limiting the concentration of free O₂ in whole blood. Then in tissue’s capillaries, O₂ diffuse into the cells depending on their metabolic activities and O₂ consumption’s rate. Thus, the concentration of O₂ in blood decreases, allowing the release of O₂ by hemoglobin according to the metabolic demand and local factors (such as pH through the Bohr’s effect). All the purpose of the blood carrier system is to ensure sufficient O₂ supply, while limiting its toxicity. One can easily see that except from the airways, other parts of the body, especially the cells are not exposed to high O₂ concentration and that hyperoxemia is not a surrogate for hyperoxia. In hyperbaric condition, the dissolved O₂ in blood is dramatically increased so it will diffuse to the cells and increased cellular O₂ concentration beyond their needs although arteriolar vasoconstriction tends to limit it until a threshold. ATM=atmosphere; P_aO₂= Arterial blood O₂ partial pressure; P_AO₂= Alveolar O₂ partial pressure ; O₂= Dioxygen; C_aO₂=arterial content in O₂ ( where S_aO₂ is the arterial blood O₂ saturation).

At tissue level, the hemoglobin unloads O₂ molecules as the surrounding P_aO₂ decreases, in accordance to local consumption rate, phenomenon enhanced by local factors such as Bohr effect [7]. Consequently, in standard conditions of pressure and temperature, the quantity by time unit of supplemental O₂ administered to the patient (i.e., fraction of inspired O₂ (F_IO₂)) and time spent above a specific threshold (the equivalent of oxygen toxicity unit used by the divers) are the paramount variables. Indeed, the cells constitutive of the airways are directly exposed to O₂ molecules and their potential toxicity, whereas other tissues are protected by the low O₂ solubility, the O₂ carrier system and the microcirculation response to excessive arterial dissolved O₂ concentrations [8]. Previous studies demonstrated that the venoarterial difference in O₂ which represents the actual quantity of O₂ delivered to the tissue/cells was not increased -it even slightly decreased- during normobaric and hyperbaric exposure to 100% O₂ compared to room air [9, 10]. More recently a study investigating the O₂ consumption during deep hypothermic cardiopulmonary bypass during pulmonary endarterectomy surgery provides interesting insights. The authors showed that despite the increase in P_aO₂ (above 500 mmHg) during cooling (from 37 to 18 °C), the venoarterial difference in O₂ and O₂ extraction decreased (reflecting the metabolism slowdown related to hypothermia). It demonstrated that the cells used only the O₂ that they need. It is a well illustration that hyperoxemia does not mean hyperoxia [11].Also, a pilot study compared the level of ROS, according to two level of F_IO₂ (40% and 70%) in mechanically ventilated patients in a context of traumatic brain injury. The blood level of ROS was similar between the two groups, suggesting that the relationship between hyperoxemia and excessive oxidative stress is not that obvious [12].

Effect of O₂ on cerebral blood flow

The effect of hyperoxemia on cerebral blood flow has been extensively studied in animals and humans. In normobaric conditions, to breathe pure oxygen leads to a small non-significant decrease in cerebral blood flow (CBF) at constant level of arterial partial pressure of carbon dioxide (P_aCO₂) [2, 13, 14]. Furthermore, studying the own effect of O₂ on CBF cannot be properly done focusing on P_aO₂ alone but should consider other conditions affecting the CBF (i.e., P_aCO₂, pH, preserved or altered cerebral autoregulation, spontaneous breathing or mechanical ventilation).

The response to change in O₂ or CO₂ partial pressures can be blunted or exacerbated following brain injury and evolves dynamically overtime according to clinical evolution [15]. Hyperoxemic vasoconstriction is a protective mechanism whose main purpose is to limit the quantity of O₂ reaching the tissue and potentially the cells [16, 17]. This vasoconstriction is a local, microcirculatory response to the increased periarteriolar O₂ concentration [17]. Although, the mechanism is incompletely understood, the role of prostaglandins on the hyperoxemic vasoconstriction pathway seems unlikely because of the lack of effect of cyclooxygenase inhibitors on hyperoxemic vasoconstriction [17, 18]. The role of direct vasoconstrictive substances such as leukotrienes seems more convincing to explain those findings [19].

P_aO₂ impact on CBF and metabolism is the cornerstone of the physiologic rationale justifying the use of hyperbaric oxygenation (HBO₂) following traumatic brain injury, stroke or anoxic encephalopathy [18]. HBO₂ have been associated with less reperfusion injuries (when applied early) possibly due to decreased endothelial neutrophils adhesion, decreased vascular permeability and vasogenic edema, decreased of intracranial pressure and the redistribution of flow from non-ischemic to ischemic area [20]. Conversely, the paradoxal hypoxia caused by the hyperoxemia-induced vasoconstriction has not been clinically demonstrated [16].

Is O₂ harmful or beneficial in a context of acute brain injury in light of physiological knowledge?

Excessive oxidative stress may occur when flow and O₂ supply are restored after a transitory ischemia/oligemia or in case of impaired O₂ diffusion to the tissue (i.e., in case of cerebral edema). So, the timeline between the occurrence of brain injury and the quantity of O₂ administered as well as the cause of the injury is essential to correctly identify if exposure to high level of O₂ may be beneficial or harmful [20, 21]. From a physiological point of view, dramatically increasing P_aO₂ (above 225–300 mmHg) in normobaric (and normothermic) condition could be potentially harmful because of the low solubility of O₂ in blood [21, 22].

However, in specific cases HBO₂ could be an interesting therapeutic option. HBO₂ increases O₂ delivery even when flow is compromised and has been evaluated in various context of ABI [23]. Pre-clinical investigation showed that HBO₂ administered within 6 h following different model of ABI was associated with reduced apoptosis, size and severity of the lesions by the maintenance of O₂ delivery despite ischemia, as well as direct anti-inflammatory effect [20]. Neuroprotective effects through decreased blood brain barrier permeability and its dysfunction were also suggested [20].

Beyond lab models, HBO₂ has also been evaluated in human patients. In a randomized trial, treatment by HBO₂ (60 min each 8 h at 1.5 atmosphere for two weeks) implemented in the early phase of traumatic brain injury reduced mortality, particularly in the subgroup of patients with an intracranial pressure over 20 mmHg [23, 24]. Conversely, results were contrasted with HBO₂ in the acute phase of stroke, as resumed by a Cochrane Review from 2014 which conclude that evidence is insufficient to recommend its routinely use [25].

In summary, the effects of supraphysiologic exposure to O₂ on CBF and intracellular oxidative metabolism are complex and depend on the clinical context, the timing and duration of the exposure following brain injury. In our opinion, the question whether hyperoxemia is beneficial or deleterious will never find a definitive response, since it is not the right way to address it. We believed that the effect of supraphysiologic exposure to O₂ cannot be addressed in a binary way, where it is safe below a specific threshold and harmful above. Individualizing the effect of O₂ alone necessitate to closely monitor other covariables such as the cause of ABI, the timing of exposure to high O₂ concentration, the P_aCO₂ among others.

We believe that the conclusion that hyperoxemia is associated with worse outcomes and higher mortality should be tempered and contextualized. We agreed that in normobaric condition, blindly increasing P_aO₂ could be harmful as it exposes the patient to O₂ toxicity (particularly the lung) without increasing enough the dissolved O₂ to have significant advantages. Evidences deriving from HBO₂ studies and supported by physiological evidence suggest that hyperoxemia could be beneficial in several clinical context. The use of HBO₂ in context of ABI is still to be evaluated by larger prospective studies as strong clinical evidence are lacking despite encouraging physiological rationale. Result from an ongoing Phase II randomized trial (NCT02407028) evaluating the optimal modality of HBO₂ to treat acute traumatic brain injury should soon give us valuable insight.

Abbreviations

ABI

Acute brain injury

CBF

Cerebral blood flow

F_IO₂

Fraction of inspired O₂

HBO₂

Hyperbaric oxygenation

O₂

Dioxygen

P_aO₂

Partial pressure of dioxygen

P_aCO₂

Partial pressure of carbon dioxide

ROS

Reactive oxygen species

Author contributions

SD and RM wrote the first draft of the manuscript. MA and HK reviewed the proof and made substantial correction. All authors reviewed the final version of the manuscript.

Funding

Financial/Non-financial disclosure: Support was provided solely from institutional and/or departmental sources.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

SysiPh group members

Roman Mounier, Sylvain Diop, Maxime Aparicio, Ariane Roujansky, Hatem Kallel.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.