Hypoxia: developments in basic science, physiology and clinical studies

Summary

Airway management is primarily designed to avoid hypoxia, yet hypoxia remains the main ultimate cause of anaesthetic-related death and morbidity. Understanding some of the physiology of hypoxia is therefore essential as part of a ‘holistic’ approach to airway management. Furthermore, it is strategically important that national specialist societies dedicated to airway management do not only focus upon the technical aspects of airway management, but also embrace some of the relevant scientific questions. There has been a great deal of research into causation of hypoxia and the body’s natural protective mechanisms and responses to it. This enables us to think of ways in which we might manipulate the cellular and molecular responses to confer greater protection against hypoxia-induced tissue injury. This article reviews some of those aspects.

‘Hypoxia not only stops the machine, it wrecks the machinery’– attributed to J. S. Haldane, 1921.

Airway management for an anaesthetic can be divided into three phases: establishment of an airway at the start of the procedure; maintenance of the airway throughout the procedure; and re-establishment of a natural airway under the patient’s own control at the end of the procedure. For each of these, the primary purpose of airway management is the avoidance of hypoxia, yet hypoxia and hypoxia-related injury remain among the main ultimate causes of anaesthetic-related death and serious morbidity [1–4]. While much progress has been made in techniques and tools to establish an appropriate airway for a procedure, two other areas of considerable research interest are related to the physiological response to hypoxia. First is the physiological attempt to maintain adequate transport of oxygen to the tissues and second is cellular (particularly neuronal) protection against hypoxia. When airway patency fails and inadequate oxygen is transported to the lungs, the natural physiological response to hypoxia is evoked, since the body’s oxygen stores are relatively limited (as covered in another chapter in this supplement [5]). Many anaesthetic and analgesic drugs themselves impair the body’s defensive reflexes to hypoxia. These effects can outlast the period of anaesthetic exposure, supervening in the postoperative period and especially during sleep (i.e. at times when these reflexes are most needed). Degrees of hypoxic injury can therefore contribute to poor outcomes, especially in patients with pre-existing cardiovascular or respiratory disorders who were already at higher risk. We need a fuller picture of the body’s response to hypoxia, of how anaesthetic drugs and agents interfere with this response, and how we might manipulate some of these natural reflexes to confer protection on organs at risk of hypoxic injury.

Adaptations to hypoxia

Considerable research has been done on cerebral protection from ischaemia/hypoxia. During hypoxia resulting from an inadequate airway, tissue perfusion may still be present with the resulting transport of other substrates (e.g. glucose) to the tissue, and removal of metabolites (i.e. carbon dioxide (Co₂) and hydrogen ions (H⁺)) from it. While the availability of glucose provides the potential for continued production of adenosine triphosphate (ATP) through anaerobic metabolism, hyperglycaemia may be detrimental during hypoxaemia, perhaps due to intracellular acidosis [6]. There exist both immediate adaptations to severe hypoxia that may provide protection over a longer period of time (e.g. ischaemic preconditioning) and longer-term adaptations (e.g. altitude acclimatisation). There is considerable variation in hypoxia tolerance across vertebrates, but even humans manifest a remarkable adaptation to hypoxia when given enough time. It has been estimated that the arterial Po₂ in climbers at the top of Mt Everest is ∼28 mmHg [7], clearly a level that would cause significant brain injury or death if imposed acutely.

Despite research in cerebral ischaemic protection (i.e. treatments before the insult) and cerebral resuscitation (i.e. treatments after the insult) there is sparse evidence for a truly effective therapy (other than hypothermia) [8]. Bickler listed the following processes consequent upon hypoxia: (1) failure of energy balance resulting in ATP deficiency; (2) uncontrolled excitatory neurotransmitter release; (3) free-radical damage; (4) inflammation and immune system over-activation; and (5) delayed cell death [9]. It is not easy to know which of these are adaptive responses and which are signs of impending cell damage [9]. At cellular level, adaptations to hypoxia are aimed at the reduction in overall need for ATP; this helps maintain the minimal functions required for cellular integrity.

In excitable cells, ATP is primarily needed for the ion pumps that maintain sodium (Na⁺) and potassium (K⁺) concentration gradients. The freshwater turtle (Trachemys scripta) is very anoxia-tolerant and has been studied extensively as a model for the mammalian hypoxic response [10]. By decreasing ion flux (‘channel arrest’) and tipping the balance between excitatory and inhibitory neurotransmitters towards inhibitory, the turtle can be anoxic for weeks at room temperature. Studying the mechanisms by which the turtle accomplishes these metabolic changes may give us clues towards new human therapies.

Another area of active research is based on the observation that an episode of hypoxia of a degree insufficient to induce injury may induce a cascade that provides protection against a subsequent ischaemic episode and this ‘ischaemic preconditioning’ may provide protection for hours or even days [11]. The timing and magnitude of the preconditioning stimulus is important as the repetitive hypoxic episodes during sleep in obstructive sleep apnoea syndrome have, by contrast, negative rather than positive health consequences. Of particular interest to anaesthesiologists is the observation that inhalational anaesthetics can provide a somewhat similar form of preconditioning against ischaemia in the brain [12] and the heart [13] but this line of research has not yet produced any practical clinical therapies.

Volatile anaesthetics and ventilatory control

The second prong of the physiological response to hypoxia encompasses the mechanisms that increase oxygen delivery to the tissue. The acute ventilatory response to hypoxia (AHVR) is a fundamental defence mechanism and is important whenever the patient’s lungs are not being mechanically ventilated. Since many of the pharmacological agents used in the peri-operative period can have profound effects on the AHVR, it is important to understand the physiology and pharmacology of this chemoreflex.

In humans, the ventilatory chemoreflex response is initiated by the carotid bodies, whose type I (glomus) cells transduce the hypoxic signal (in a manner not yet fully established [14]). It is generally agreed that a membrane K⁺ channel is involved, but there may be several other mediators. Hypoxia in some manner causes K⁺ efflux from type I cells, resulting in membrane depolarisation, which leads to calcium (Ca²⁺) influx. This causes vesicles containing neurotransmitter (there are several candidates for the primary transmitter involved) to fuse with the membrane; the neurotransmitter then excites the afferent nerve terminals of the carotid sinus branch of the glossopharyngeal nerve, whose activity is increased. Feedback to the medulla via these nerves evokes a reflex efferent increase in minute ventilation.

The carotid bodies are essential for this response. The human brain does not respond with any excitation to hypoxia; rather, brain neuronal activity is reduced, in part due to protective mechanisms to reduce oxygen consumption [15].

The observations of Knill et al. in the 1970s (since confirmed by many other studies) established that in humans, volatile anaesthetics significantly depress the AHVR [16–18]. At low dose (< 0.2 minimum alveolar concentration(MAC)) the degree of depression is ∼50%; at higher doses of ∼1 MAC the hypoxic response is virtually abolished. Since low doses of agent persist for many hours after surgery, this depression of an important protective reflex is of concern. To adapt Knill’s phrase about the carotid bodies after anaesthesia: ‘the watchdogs are asleep’ and not protecting us from hypoxia. One interesting finding has been that some agents (e.g. isoflurane and sevoflurane) obtund the chemoreflex much less than others (e.g. halothane) [19–24].

A possible mechanism by which agents inhibit the reflex is through interaction with a specific background K⁺ (K_B) channel in the carotid body [25]. Since we now know that certain anaesthetic drugs have less action upon this channel than in others [26, 27], it offers potential to design drugs without the side effect of inhibiting the hypoxic chemoreflex.

Notwithstanding the fact that the background K⁺ channel is a key candidate for anaesthetic effect at carotid body level, there are other suggested mechanisms by which hypoxia is detected. Since the mechanism of transduction by definition confers protection against hypoxia for the carotid body cells, manipulating this mechanism (if it were clearly identified) could protect other organs from hypoxic injury. Specific mechanisms (other than the K_B channel) postulated to be involved in hypoxic transduction include reactive oxygen species, mitochondria, hypoxia-inducible factors and haemoxygenases. These are reviewed in brief below.

ROS mechanism

Reactive oxygen species (ROS) are formed in hypoxia and could alter the redox status of signalling molecules and the function of membrane ion channels [28, 29] (i.e. this mechanism is not necessarily entirely independent of some of the mechanisms described below). Two ROS-producing sites proposed as oxygen sensors are the reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase and mitochondria systems. In an elegant series of experiments, Dahan et al. posed the possibility that halothane might reduce the carotid body/ventilatory response to hypoxia by producing ROS in an hypoxic environment, antagonizing the effect of hypoxia on K⁺ channels. Consistent with their supposition, experimental pretreatment of human subjects with an antioxidant cocktail of ascorbic acid and α-tocopherol reversed halothane and isoflurane inhibition of the hypoxic response [28, 29]. This is a potentially important development as it raises the possibility that antioxidants could be used to preserve the hypoxic ventilatory response after anaesthesia.

Mitochondrial mechanism

Mitochondria uncouplers raise cytosolic Ca²⁺and reduce background K⁺ permeability in glomus cells [30]. In hypoxia, mitochondria generate signals that alter membrane ionic conductances (e.g. through modification of the cell redox status or via reduction of cytosolic ATP). Mutations in the mitochondrial complex II (particularly in the small membrane-anchoring subunit of succinate dehydrogenase (SDHD)) are the main cause of familial hereditary paraganglioma (PGL), a vascular, catecholamine-secreting tumour [30]. As PGLs display cell changes similar to those seen after chronic hypoxaemia it can be argued that the cause of tumourigenesis is a defect in sensing environmental oxygen levels [31].

Prolyl/asparagyl hydroxylases and HIF pathway

Prolyl/asparagyl hydroxylases use oxygen (together with ferrous ions (Fe²⁺) and α-ketoglutarate as co-substrates) to hydroxylate specific proline/asparagine residues of hypoxia-inducible transcription factors (HIF)-1α, and its isoforms to regulate the expression of numerous hypoxia-sensitive genes. In hypoxic conditions, the lack of hydroxyl groups in proline and asparagines residues stabilises the HIF molecule [32]. Hydroxylation of HIF in oxygen occurs rapidly, so it is possible that this mechanism could be involved in acute responses to hypoxia.

Haemoxygenase mechanism

Haemoxygenase (HO-2) is a ubiquitous antioxidant enzyme that uses oxygen to convert haem into biliverdin, iron and carbon monoxide (CO) [33]. HO-2 combines with specific membrane K⁺ channels to modulate their oxygen sensitivity. HO-2 is expressed in rat carotid body glomus cells; in addition, K⁺ channels recorded in patches excised from these cells are activated by HO-2 substrates (haem and NADPH). Therefore, it is suggested that HO-2 could act as an ‘oxygen sensor’ through the production of CO, which is by itself a K⁺ channel activator.

Opioids and respiratory control

It is well established that morphine and its main metabolite, morphine-6-glucuronide (M6G) depress ventilation and also the ventilatory responses to hypoxia and hypercapnia [34]. Mice lacking the μ-receptor (exon-2 μ-opioid receptor, MOR-1 gene knockout mice) have no antinociceptive or respiratory depression response to either drug. MOR-1 mice also display higher resting respiratory frequency and an increased frequency response to hypercapnia [35]. In humans and animals, morphine and M6G are more potent analgesics and respiratory depressants in women than in men, possibly related to the developmental and organisational effects of sex hormones in prenatal or early postnatal life, or related to polymorphisms in μ-receptors between men and women [36].

Interestingly, the effects of M6G on ventilation and analgesia are dissociated, needing a higher plasma concentration to cause analgesia than respiratory depression (with morphine, analgesia and respiratory depression go hand-in-hand). Furthermore, M6G selectively depresses baseline ventilation as opposed to hypoxic ventilation (morphine by contrast selectively depresses hypoxic ventilation). These effects may be related to polymorphisms of the μ-receptor (particularly the A118G polymorphism) [37]. What these findings show is that it may not only be theoretically possible to design a drug with analgesic effects, but also one which preserves the hypoxic response.

Response-surface modelling of opioid effects on ventilation allows three-dimensional plots of opioid concentration, minute ventilation (or ventilatory response to CO₂) and anaesthetic concentration. This reveals that opioid-anaesthetic interactions on ventilation are additive at low concentrations but synergistic at high concentrations. Rapid bolus injections can lead to apnoea, suggesting that in the context of airway management, carefully titrated infusions rather than intermittent boluses are desirable (e.g. sedation for fibreoptic intubation techniques or postoperatively for patient-controlled analgesia in at-risk patients) [38].

In contrast to general anaesthetics, whose respiratory effects appear largely located in the peripheral chemoreceptor, the effects of opioids are primarily central, most likely on the pre-Bötzinger complex of the brainstem (or possibly the Kolliker-Fuse nucleus). This central action underlines the influence of background levels of arousal on respiratory responses (i.e. sleep can enhance whereas arousal can antagonise the respiratory depressive effects) [34].

Drugs and collapse of the upper airway

Induction of anaesthesia leads, to a varying degree, to collapse of the upper airway. Indeed, ‘difficulty’ in airway management is often taken to include difficulty in maintaining patency of the upper airway and allowing manual bag-mask ventilation of the lungs. Certain patient groups are prone to collapsibility of the upper airway in these circumstances (e.g. the obese, pregnant women and those with genetic abnormalities such as Down’s or Treacher-Collins syndrome, muscular dystrophies, etc) [39].

Collapse of the airway occurs at narrower and non-rigid sites, namely the tonsillar area, the velopharynx and the retrolingual region. The upper airway may be envisaged even more simply as a collapsible segment (pharynx) between two rigid segments (the nose and the trachea). When tissue pressure around the pharynx for whatever reason exceeds the pressures within the nasal and tracheal segments, collapse ensues. This can to some extent be overcome by increasing upstream pressure (e.g. continuous positive airway pressure). A centrally mediated drive, both tonic and phasic, from the brainstem helps maintain airway patency by activating the dilator muscles. This is naturally reduced during anaesthesia, and can be restored by arousal. Patency is enhanced by local upper airway reflexes. Pressure gradients in the pharynx act as the stimulus for the reflex, which is why topical anaesthesia of the pharynx can induce airway obstruction [39].

General anaesthetics have a dose-dependent depressive effect on upper airway muscle tone, and a greater effect on the diaphragmatic tone. Thus, as the upper airways collapse there is inspiratory flow limitation (as downstream pressure declines during inspiration).

Neuromuscular blocking drugs are of course potent inhibitors of upper airway tone. Complete blockade necessitates artificial ventilation, and the anaesthetist is always present during this period. The real concern, however, is incomplete reversal of blockade, when the level of patient supervision is reduced in the recovery area (or even in the ward). Incomplete recovery is most likely to affect more sensitive muscle groups, which include the pharyngeal muscle and oesophageal sphincters. The resulting sensations can be very distressing for patients and lead to a vicious cycle of dyspnoea, increased oxygen consumption, but (because of upper airway collapse) inspiratory flow limitation. Furthermore, aspiration can occur: the Fourth National Audit Project of the Royal College of Anaesthesists and Difficult Airway Society (NAP4) in the UK has shown that overall, this was a leading cause of hypoxia causing anaesthesia-related mortality [2–4].

Non-depolarising drugs can directly inhibit the carotid body (acetylcholine is a putative neurotransmitter in type I cells acting on nicotinic receptors on the afferent nerve terminal) and this obtunds the hypoxic ventilatory response (synergised by residual volatile agents and/or opioids) [40]. Yet there remains a dilemma as to how best to reverse the action of neuromuscular blocking drugs. If neostigmine is given when paralysis is too profound, it may not reverse blockade. If given too late when paralysis is almost fully recovered, neostigmine may itself weaken muscle power. Sugammadex offers an opportunity for safer reversal of paralysis, but its precise role in clinical algorithms is yet to be established [41].

For all these reasons, there is considerable opportunity for research into areas like optimal monitoring of neuromuscular blockade (especially in the relevant muscle groups such as the pharynx), and in the potential interactions of neuromuscular blocking drugs with other anaesthetic drugs and reversal agents.

Hypoxaemia during recovery from anaesthesia

In the third phase of airway management, the transition to the patient’s own breathing, many factors are involved: alveolar hypoventilation; atelectasis; blunting of hypoxic pulmonary vasoconstriction; blunting of the hypoxic chemoreceptor response; decreased arousal response and decreased upper airway tone. Individually summarised above, these act synergistically to increase the risk of hypoxaemia during this critical period. Complicating this scenario (as mentioned above), is the fact that most patients move at this time to an area of lesser supervision than when under the direct care of the anaesthesiologist.

Keeping patients awake, incentive spirometry and maintenance of pulmonary toilet are key ways to maintain oxygenation during this period. However, since patients also need to sleep, the interplay of sleep and sedation, analgesia and anaesthesia is important.

The first 72 h after surgery is a period of high risk (one in three cases in NAP4 occurred at emergency or recovery [2–4]). A prospective survey of > 18 000 patients in the post-anaesthesia care unit revealed a complication rate of ∼7% related to airway obstruction [42]. Catley et al. found that use of intravenous morphine resulted in greater oxygen desaturation than that in regional anaesthesia alone [43]. Knill et al. previously identified that rebound rapid eye movement (REM) sleep occurring during a week after abdominal surgery contributed to periodic breathing and hypoxaemia [44]. Sleep-related breathing disturbances are exaggerated by sedation, anaesthesia and the coexistence of factors such as obesity and obstructive sleep apnoea (OSA). Thus, even a mild case of OSA can develop into a severe case of postoperative apnoea leading to greater morbidity [45]. Recently, Liao et al. verified in a retrospective analysis that patients with OSA had an increased risk of postoperative hypoxia compared with controls, and this effect was more prominent on the hospital ward rather than in the post-anaesthesia care unit [46]. Ramachandran et al., in reviewing a large database, found that the incidence of unanticipated postoperative tracheal re-intubation in patients with OSA was ∼1% within the first three postoperative days [47, 48]. A caveat is that although the presence of OSA predicts worse postoperative outcomes, with respect to respiratory compromise, its severity does not necessarily correlate with poor outcome [49].

The key message is that ‘airway management’ does not stop with the removal of the airway device at the end of the surgical procedure. Continuous oxygen and ventilation (i.e. capnography) monitoring for all patients would be ideal, but its cost is as yet unjustified and early identification of at-risk patients is warranted [50, 51]. In this group, measures such as tailored anaesthetic technique (e.g. avoidance of opioids to the extent possible, full reversal of neuromuscular blockade, and the use of regional anaesthesia) are all ways to help prevent hypoxaemia from upper airway collapse and hypoventilation. The use of continuous positive airways pressure (CPAP) should continue into the postoperative period in those patients who need it at home. While Ramachandran et al. found that 83% of home CPAP users brought their machines into hospital [48], Liao et al. found that only 49% actually received CPAP postoperatively [46]. Drummond et al. have questioned the utility of ‘routine’ postoperative CPAP on the grounds that pre-operative CPAP settings may not be adequate in the postoperative setting, and respiratory therapists may be needed to titrate the appropriate CPAP level [51, 52]. Importantly, though, Rennotte et al. reported a reduction in postoperative complications in patients who used CPAP therapy pre-operatively, upon extubation, and continuously 24–48 h after surgery [52]. More recently, Squadrone et al. found postoperative CPAP to reduce re-intubation rates, length of stay in the intensive care unit, incidence of postoperative pneumonia, infection, sepsis and death [53]. While supplemental oxygen alone may prove beneficial for many patients, it may increase the incidence and duration of apnoeic episodes in OSA patients [54].

Conclusions

Fully understanding ‘difficult airway management’ requires us to go beyond just technical aspects of equipment such as supraglottic airways or tracheal intubation aids. It is also important to appreciate – at fundamental cellular and molecular levels – how drugs such as general anaesthetics, opioids and neuromuscular blocking agents impair the body’s defence mechanisms against hypoxia. These defences include the hypoxic chemoreflex and also patency of the upper airway. Establishing why some patients are more susceptible than others to these drug effects, or developing drugs or drug combinations with lesser depressive actions on these protective reflexes, will be beneficial. Holistically, managing a difficult airway (or indeed any airway) is an issue of preventing hypoxia and its adverse consequences. Continuous positive airway pressure remains the gold standard therapy for maintenance of upper airway patency during sleep. Identifying patients at risk of sedation-related breathing disturbance and initiating CPAP in the postoperative period in at-risk patients is one step to prevent postoperative hypoxaemia.

Competing interests

No external funding or competing interests declared.