1. Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a novel member of the enveloped ribonucleic acid (RNA) beta coronavirus family, is the infectious agent causing the coronavirus disease 2019 (COVID-19) pandemic that has resulted in significant health, social, and economic burdens worldwide [1]. Severe COVID-19 disease is associated with male gender predominance, older age, metabolic syndrome, and obesity [2]. Moreover, some patients develop a multi-system disease process involving major organs with acute myocardial injury, acute kidney injury, haematological abnormalities, and intracerebral complications with prothrombotic tendencies [3]. A minority of hospitalised patients with severe COVID-19 pneumonia develop acute hypoxaemic respiratory failure (AHRF) necessitating intensive care admission and the initiation of mechanical ventilation to support adequate arterial oxygenation [4]. As the management of these critically ill patients continues to evolve, the pathophysiology of severe COVID-19 lung injury remains an intriguing phenomenon [5]. Although COVID-19 is defined as a single disease entity with a single causative agent, diverse clinical phenotypes may warrant individualised treatment approaches. These phenotypes are likely to reflect the complex host–virus interaction associated with SARS-CoV-2 infection, particularly the degree of host immunothrombotic response during and after the viral illness [6]. There is often a lag of 7–12 days between infection and the development of AHRF with SARS-CoV-2 infection, and antiviral strategies at this stage appear to offer no survival advantage [7][8]. Some patients go on to develop sustained hypoxaemic respiratory failure with prolonged hospitalisations.
The consequences of hyperoxic organ toxicity have been evaluated in many animal studies since the 19th century. These studies present the organ-specific deleterious effects of hyperoxia in different animal models and experimental conditions with variations in the FiO
2, barometric pressure, and duration of exposure. Although there were variations in tolerance, pulmonary toxicity was consistently reported with characteristic pathological changes
[9]. The studies investigating the effect of alveolar hyperoxia on human lungs are primarily conducted on healthy humans, with non-injured lungs. As a result, the exact dose and duration of oxygen exposure for human pulmonary toxicity and lethality are largely unknown, especially in injured lung conditions such as viral pneumonia.
2. Ubiquitous Use of Oxygen May Be a Problem—Pre COVID-19 Studies
The duration of tolerance and the lethal dose of oxygen have been evaluated by several small and large animal studies of hyperoxia over the past two centuries
[9][10]. There were considerable variations in the tolerance of pulmonary toxicity at various oxygen tensions and survival ability between animal species; mostly, high dose and prolonged exposure of an FiO
2 between 0.90 and 1.0 are associated with acute respiratory failure and death
[9]. The toxicity appears to be proportionate to the FiO
2 [10]. When challenged with an FiO
2 of 0.85–1.0, animal models of lower primates (rhesus monkeys, baboons, sooty mangabeys, and squirrel monkeys) fared better and survived longer, averaging around 8 days (range 3–17 days), than other small animals (rats, mice, guinea pigs, and birds)
[9][11]. Moreover, susceptibility and the magnitude of pulmonary oxygen toxicity are variable even between individuals from the same species, suggesting an individual genetic predisposition
[12]. Animal models also suggest that there is age-related differences in response to hyperoxia, where neonatal lung is more resistant to hyperoxia-induced lung injury compared with adult lungs, implying that the progressive development of innate immunity may contribute to hyperoxia-induced lung injury
[13]. Pathologically, hyperoxic lung injury is characterised by diffuse alveolar epithelial and endothelial damage with exudative pulmonary oedema and capillary leakage very similar to the characteristic features of ARDS
[14][15].
Although human hyperoxic challenge studies for prolonged periods are no longer ethically feasible, historical work demonstrated normobaric exposure to an FiO
2 of 1.0 for 14 h led to substernal distress and pleuritic chest pain
[16]. In a later study, a longer duration of exposure (30–74 h) resulted in the development of cough and progressive dyspnoea with an associated decline in vital capacity and diffusing capacity for carbon monoxide (DLCO)
[17]. Abnormalities in tracheal mucociliary movement are evident after 3 h of exposure to an FiO
2 of 0.90–0.95
[18]. Moreover, an FiO
2 of more than 0.95 for 17 h leads to significant alveolar capillary leak with increased mediators of fibroblast recruitment and proliferation in healthy adults
[19]. These limited numbers or normobaric hyperoxic human studies combined with translation from large primate and small animal studies support the potential for pulmonary toxicity and lethality in a normal uninjured lung, which in general, is associated with hyperoxia of an FiO
2 > 0.70 beyond an exposure duration of 24 h
[9][11]. However, more importantly, infected injured lungs may respond differently to hyperoxic challenges than a normal lung. Indeed, hyperoxia in legionella pneumonia increased lethality with accelerated apoptosis in rodent models
[20]. The implications of combined insults of viral pneumonia and hyperoxia in the development and progression of acute lung injury is largely unknown.
Several oxygen intervention studies of mechanically ventilated or intensive care and critically ill patients investigating various oxygen targets of hyperoxaemia have been published to date. A systematic review of oxygen therapy of >16,000 hospitalised adult patients with acute illness (IOTA) concluded that liberal peripheral oxygen saturation (SpO
2) targets beyond 94–96% are associated with increased in-hospital morality
[21]. Despite this evidence, recent randomised controlled trials of patients with AHRF (HOT-ICU) and ARDS (LOCO2) suggest no clinical benefits from conservative arterial oxygen (PaO
2) targets between 50 and 70 mmHg
[22][23]. Further, larger studies are currently underway nationally, UK-ROX, and internationally, MEGA-ROX, indicating that this is an important research question with ongoing controversy
[24][25][26]. Although these trials aim to accept a degree of permissive hypoxemia, they do not address the negative impact of alveolar hyperoxia in patients with severe hypoxemic respiratory failure requiring high fractional inspired oxygen.
3. Oxygen Toxicity Mechanisms
The cellular pathways leading to hyperoxia-mediated lung injury are complex and beyond the scope of this research. Briefly, lungs are vulnerable to oxidative damage, which is exacerbated during inflammatory conditions. Hyperoxia disrupts the normal physiological homeostatic balance and increases oxidative stress by inducing highly reactive mitochondrial oxidative stress mediators such as superoxide (O
2−), hydrogen peroxide (H
2O
2), hydroxyl radicals (OH
−), and peroxynitrite anions (ONOO
−). Unopposed reactive oxygen species (ROS) lead to compromised cellular function with a predisposition of oxidative damage to deoxyribonucleic acid (DNA) material, lipids, and proteins
[27]. Viral infections are also associated with increased oxidative stress
[28]. Furthermore, imbalances in antioxidant mechanisms including the enzymes superoxide dismutase (SOD), catalase, and glutathione peroxidase and small defence molecules such as glutathione, ascorbic acid, and vitamin E, may result in the presence of increased oxidative stress mediators leading to further direct mitochondrial and cellular damage
[29][30]. A deficiency in native plasma antioxidants is a common feature in COVID-19 infection and COVID-19 critical illness
[31][32][33]. In healthy physiological states, the balance of oxidant and antioxidant is highly regulated and alterations in this equilibrium can lead to a proinflammatory state with a subsequent influx of inflammatory cells, the activation of cytokine cascades, and increased vascular permeability
[34]. Alveolar hyperoxia with reduced native antioxidants in severe SARS-CoV-2 infection is likely to contribute to an overwhelming redox imbalance (
Figure 1).
Figure 1. Increased oxidative stress as a result of a combination of SARS-CoV-2 viral infection and alveolar hyperoxia. * Reactive oxygen species.
Molecular pathways leading to hyperoxic acute lung injury involves the activation of a multitude of signal transduction pathways of cellular homeostasis. In healthy physiological states, there is a balanced regulation of cell growth and cell death by several regulatory mechanisms of apoptosis and necrosis. Exposure to hyperoxia and the subsequent generation of ROS by nicotinamide adenine dinucleotide phosphate oxidase (NOX) 1 phosphorylation leads to changes in protein kinases, transcription factors and cellular apoptotic/necrotic pathways. Mitogen-activated protein kinase (MAPK) signalling cascades involving extracellular signal-regulated kinases (ERK1/2), c-Jun N-terminal kinase (JNK), and p38 kinase are all implicated in hyperoxic acute lung injury
[35]. The role of these stress-activated protein kinases in both the upregulation and downregulation of hyperoxia-induced cell death has been studied extensively in various animal models of cellular hyperoxia
[36]. Moreover, the hyperoxic exposure of lung stimulates molecules involved in the regulation of cell death, Fas and the Fas ligand with downstream activation of Caspase-8, and pro-apoptotic proteins (Bax, Bid, Bim, and Bak). A subsequent increase in protein kinase C delta type (PKC-δ) leads to the release of mitochondrial cytochrome C, and further cleavage of caspase-3 and 9 results in apoptotic and necrotic cell death
[37]. Other transcription factors such as nuclear factor kappa B (NFκB), activator protein-1 (AP-1), signal transducer and activator of transcription (STAT), nuclear factor-erythroid 2-related factor 2 (Nrf2) and Toll-like receptor 4 (TLR4) play intricate counterbalanced roles in hyperoxic acute lung injury. The protective role of Nrf2 in antioxidant defence during hyperoxia is well established, and genetic polymorphisms in the Nrf2 gene may increase the epigenetic susceptibility to developing hyperoxic acute lung injury
[38][39].
4. Alveolar Hyperoxia Induced Surfactant Damage
Pulmonary surfactant is a mixture of lipids and proteins at the air–liquid interface that minimises surface tension forces and prevents alveolar collapse. A surfactant is synthesised and metabolised by AT-II cells. Surfactant composition varies between animal species, but in humans, 70–80% of phospholipids are phosphatidylcholine (PC), with dipalmitoyl phosphatidylcholine (DPPC) being the major PC with the functional ability to reduce surface tension
[40]. Surfactant deficiency is a commonly recognised feature in patients with ARDS, with variations in synthesis, metabolism, and functional inhibition from alveolar inflammatory milieu
[41]. In vitro studies suggests that the surface activity of surfactant is impaired when directly exposed to oxidation
[42]. A pulmonary surfactant exposed to oxygen free radicals can lead to the direct inactivation of phospholipids and proteins
[43]. Moreover, in vivo animal studies showed that prolonged exposure to hyperoxia results in increased levels of oxidative stress with associated alterations in surfactant metabolism and turnover with reductions in lung antioxidant levels
[44][45][46].
An intact surfactant system is fundamentally required for alveolar stability. SARS-CoV-2 pneumonia presents a double-edged sword, causing impairment to the surfactant system by AT-II cell death in combination with oxidative damage from alveolar hyperoxia
[47]. Changes in AT-II cell proliferation and apoptosis are a common pathological feature of SARS-CoV-2 pneumonia. Alveolar epithelial cellular invasion by SARS-CoV-2 via ACE2 surface receptors causes increased AT-II cell apoptosis and can theoretically impair surfactant synthesis, secretion, metabolism, and recycling of functional pulmonary surfactant. However, the impact of SARS-CoV-2 infection on the surfactant system has yet to be explored. Alveolar hyperoxia may also compromise surfactant metabolism and the functional ability of surfactant to reduce surface tension resulting in poor lung compliance and worsening hypoxemia
[48][49][50]. A reduction in overall surfactant pool size and functional ability to reduce surface tension is compromised during ARDS, and the same principle may apply during COVID-19 pneumonia
[51].
Exogenous surfactant therapy has proved ineffective in adult patients with ARDS. A meta-analysis of exogenous surfactant replacement in ARDS patients demonstrated no survival benefit but some improvement in oxygenation within the first 24 h, which was not sustained
[52]. This implies that the beneficial effect of a surfactant is possibly short-lived and a longer duration of therapy until recovery may need to be considered
[53]. Moreover, a supplemented exogenous surfactant will likely face the same fate as an endogenous surfactant, with increased oxidative damage leading to a poor functional ability to maintain a low surface tension. An in vitro study of bovine surfactant suggested that surfactant performance is severely compromised when exposed to high concentrations of oxygen
[51]. However, the effect of high FiO
2 on surfactant composition, metabolism and functional surface tension-reducing ability in vivo, particularly in human injured models, has not been studied. Animal models of surfactant supplementation following hyperoxic exposure show an attenuation of alveolar oxygen toxicity, lung injury, and alveolar capillary permeability
[54][55]. Surfactant also has antioxidant properties, and administration reduces the oxidative stress in animal models of hyperoxia
[56][57][58]. From these limited available studies, it is tempting to speculate that an exogenous surfactant may be used as an antioxidant to minimise oxygen toxicity or in combination with other antioxidants to improve the surface-active properties of both endogenous and supplemented exogenous surfactants.
5. Alveolar Hyperoxia and the Expression of ACE2 Receptors
It has been postulated that the low incidence of COVID-19 among high-altitude inhabitants is possibly due to the downregulation of ACE2 expression during chronic hypoxia leading to fewer available receptors for viral entry
[59]. However, the interaction between ACE2 expression and the risk of COVID-19 infection, progression into critical illness, and recovery is likely to be complex. Moreover, there are contradicting hypotheses as both hypoxia and hyperoxia seem to modify ACE2 expression in alveolar epithelia
[60]. As detailed in an earlier section, the maintenance of normal ACE2 expression is crucially important for host immune response, and elevated levels of ANGII are associated with increased SARS-CoV-2 viral load and the severity of lung injury
[61]. While increased ACE2 expression may increase the risk of viral infection, reduced ACE2 receptor availability can result in increased levels of ANGII and pulmonary fibrosis
[62][63]. Infection with SARS-CoV-2 depletes host ACE2 and may contribute to the detrimental effects seen in the respiratory system. ACE2 protects against fibrosis, and in ACE2 knockout animal models, there is eventual development of severe lung disease and fibrosis
[64][65]. An adult mice model of hyperoxia was associated with decreased lung ACE2 expression and increased ANGII/ANG (1-7) ratio, and co-treatment with ACE2 agonists mitigated the oxidative stress
[66]. Likewise, hyperoxia downregulates ACE2 in human foetal fibroblasts, which may explain the development of BPD in neonatal lung disease
[67]. However, the degree and duration of hyperoxia required to induce in vivo ACE2 modifications in humans is not known. All these findings suggest that a combination of SARS-CoV-2 infection and hyperoxia are likely to contribute to an overwhelming downregulation of ACE2 expression and increasing ANGII levels, leading to a potentially lethal lung disease with eventual fibrosis. However, the multiplatform REMAP-CAP trial in critically ill patients recently concluded that neither ACE inhibition nor direct angiotensin receptor block improve clinical outcomes
[68].
6. Alveolar Hyperoxia Induced Pulmonary Vascular Changes
ACE2 receptors are also expressed by pulmonary vascular endothelial cells. Alveolar capillary endothelial invasion and subsequent vascular damage with changes in cellular morphology and apoptosis are suggestive of endothelial inflammation as a prominent aetiology for pulmonary organ dysfunction precipitating respiratory failure and death
[69][70]. The morphological changes in hyperoxic acute lung injury are also associated with pulmonary vascular changes, with damage to the pulmonary capillary bed and capillary leak resulting in pulmonary oedema. Although alveolar hypoxia increases pulmonary vasoconstriction and pulmonary vascular resistance (PVR), and conversely, hyperoxia leads to pulmonary vasodilation and a reduction in PVR, prolonged periods of hyperoxia may lead to a blunted vasodilatory response to nitric oxide (NO). Moreover, prolonged hyperoxia may also cause oxidative stress, which impairs the vasodilatory effects of endogenous and inhaled nitric oxide (iNO), potentially altering the potential for vascular reactivity.
7. Hyperoxia Induced Immune Dysfunction and Secondary Bacterial Infections
In critically ill patients, hyperoxaemia is independently associated with ventilator-associated pneumonia
[71]. Secondary bacterial and ventilator-associated pneumonia is relatively common in patients who are ventilated for COVID-19 pneumonia
[72]. The commonly identified organisms are klebsiella pneumonia, pseudomonas, and staph aureus species
[73]. Animal models of in vitro cultured macrophages when exposed to hyperoxia demonstrate compromised macrophage-driven innate immune functions
[74][75]. Furthermore, hyperoxaemia is associated with an increased susceptibility to Gram-negative bacterial infection with increased mortality
[76]. Moreover, the lung and gut microbiomes are greatly alerted in animal models of alveolar hyperoxia and in critically ill patients receiving high concentrations of oxygen therapy, predisposing to secondary bacterial infections
[77][78]. The hyperoxia-induced impairment in the innate immunity may in part explain the increased incidence of nosocomial infections seen in mechanically ventilated COVID-19 patients.