The topic of gas emboli is miscellaneous enough to be fitted into the "misc" section of SAQs. The most likely causes include surgical mishaps, particularly cardiothoracic (and so it could have easily slotted into the cardiothoracic ICU section). If discussion were limited to cerebral arterial gas embolism, it could be classified as a predominantly neurological issue and parked in the neurology/neurosurgery section. These would be logical options. However the author of Deranged Physiology is unencumbered by logic, and had instead opted to put gas emboli into the Trauma/ Drowning section, as by far the most interesting cause of such emboli is decompression injury, and because it allows him to digress extensively on the topic of exposing humans to pressurised gas.
This has only ever come up once in the exams, in Question 21 from the first paper of 2017. Specifically, the college wanted to hear about the trials and tribulations of transporting a patient with a cerebral arterial gas embolism - how the specific pathology affects aeromedical retrieval, for example. With any luck, this topic (like traumatic asphyxia and resuscitative thoracotomy) will be the subject of a once-off "everybody fails" question, never appearing again in the exams. One questions the role of examining such esoterica. Did this question really discriminate safe competent intensivists from unsafe ones?
Whinge aside, let us explore the literature references. The most often cited article for this topic is the NEJM paper by Muth and Shank (2000), available for free through dive-shield.us. It is unclear whether these guys are aware that they are hosting it. For a detailed discussion of venous gas emboli, the single best resource is probably Palmon et al (1997). To answer Question 21 with a question-specific article, one would use Jeffrey Stephenson's 2009 paper, "Pathophysiology, treatment and aeromedical retrieval of SCUBA-related DCI"
In brief summary:
Clinical features of gas embolism:
- PEA arrest
- "mill wheel" murmur
- Gas bubbles in the retina (on ophthalmoscopy)
- Drop in EtCO2
- ECG features of right heart strain
- Pulmonary hypertension
Management of gas embolism:
- Avoid giving N2O
- Increase FiO2 to 100%
- Put the patient in supine position
- Try to aspirate the gas from the RV using a long catheter, eg. a 25cm CVC
- Hyperbaric oxygen therapy is indicated
- Antiepileptic therapy is indicated if the patient has a cerebral gas embolism
Causes of gas embolism
Without further ado, here are the major reasons for gas emboli.
Mixed (i.e. possibility of both arteral and venous)
In order for gas to make its way into your circulatory system, several things need to go wrong in a major way. Usually, there needs to be a breach in the circulatory system (a wound, a vascular access device, etc) and the blood in this open vessel needs to be under lower pressure than the available gas. If not for the latter clause, all minor paper cuts would give rise to air emboli. These preconditions mean that venous emboli are by far more likely (as a part of the upper body venous circulation is constantly under negative pressure) and that any invasive procedure involving insufflation poses a potential risk (and this includes positive pressure ventilation). The more rare (and interesting) scenario is one where abrupt ambient pressure changes affect the solubility of dissolved gases in the body fluids, causing them to emerge out of solution in the form of bubbles. A possible third option is isobaric counterdiffusion, where gases with different blood solubility compete for limited "space" within the liquid phase, the more soluble gas forcing the less soluble one out of solution. This is generally limited to the skin and middle ear, and (though painful and crippling) is rarely a reason for ICU admission.
Massive venous gas embolism as the cause of cardiac arrest
How much air do you need to kill somebody? Basically, you need to fill the right ventricular outflow tract with gas to cause a cardiac arrest (it usually looks like PEA with tachycardia). Oppenheimer et al (1953) observed this phenomenon in dogs, and found that 7.5ml/kg was the LD100 (i.e. every dog died). In contrast, slow infusion of gas seems relatively benign; Hybels (1980) reports that up to 1400ml of gas is reasonably well tolerated in dogs, provided it happens over several hours.
What about humans? Palmon et al (1997) give a good rundown of these events, and offers a bibliography of case reports. For instance, Allen Yeakel (1968) reports on a lethal venous air embolism of approximately 136ml, which was administered accidentally during an exchange of a plastic blood storage container (the patient was undergoing a carotid endarterectomy). Harrison Martland in 1945 wrote an article describing fatal air emboli in patients who used "powder insufflators" to squirt air into their vaginas for the purpose of antitrichomonal treatment, a technique which is thankfully no longer a part of the gynaecological repertoire because the risk of air embolism is now well-appreciated. That has not stopped people from insufflating each other's vaginas recreationally, with fatal consequences. Judging by the number of case reports this is actually suprisingly commonplace, which somewhat redefines the term "high-risk sexual activity". Less hideous case reports are also available where emboli are the consequence of lung biopsy, bronchogenic cyst rupture, gastrointestinal endoscopy and arthroscopy. Given that it is hard to estimate the volume of air postmortem, Palmon et al conclude that the sudden injection of about 100-300ml of air is enough to cause circulatory arrest.
Clinical features and complications of sub-lethal gas embolism
Obviously, the exact pathophysiological features of gas embolism depend heavily upon where precisely the embolism went. The cerebral gas embolism in Question 21 from the first paper of 2017. Specifically, the college wanted to hear about the trials and tribulations of transporting a patient with a cerebral arterial gas embolism. The following discussion is of the generic clinical consequences of a venous gas embolism, ranging from symptoms and signs to TTE findings and PA catheter measurements.
- Tachypnoea and dyspnoea: If the collection of emboli in the pulmonary circulation was slow, when about 10% of the pulmonary circulation is filled with gas it tends to produce what Palmon et al call a "gasp reflex", i.e. the sensation of breathlessness. This can make things worse (eg. each deep panic-driven breath entrains more gas into the carelessly unclamped CVC lumen).
- Hypoxia due to shunt: There is usually some hypoxia associated with a venous gas embolism, and this is due to some sort of worsened shunt. Blood does not flow through the air-filled pulmonary vessels; ergo all blood flow is directed into airless vessels, and there may not be enough time to oxygenate all of it on the way though the remaining functional pulmonary circulation. In short, a V/Q mismatch develops where Q is the problem.
- Shunt-related changes in the end-tidal CO2: That is to say, just as in the case of pulmonary embolism there will be an abrupt drop in EtCO2, resulting from the sudden increase in pulmonary dead space.
- Cardiac murmur: A characteristic "mill-wheel" murmur may be audible, but is a late sign - by the time your patient is mill-wheeling they would be peri-arrest.
- Gas bubbles in the retina, on fundocscopy
- ECG changes associated with gas embolism: There is no pathognomonic wave here, but ECG will be different-looking. According to a case report by Cooney et al (2011), changes may include tachyarrhythmias, AV block, right ventricular strain (right axis deviation, T-wave inversion and ST depression in anterior and inferior leads), poor R-wave progression and ST segment changes.
- Increased pulmonary arterial pressures: It would seem strange that the right ventricle might experience increased resistance given that the obstruction to flow consists of nice soft air, but indeed the PA pressure tends to rise.
- Paradoxical embolisation: This can happen when the right sided pressure gets too high. Either a previously dormant foramen will open, or the air will penetrate into the left sided ot the circulation via pulmonary capillaries. Either way, systemic arterial emboli will result. Apart from the brain, probably the next most horrible site of embolisation would be the coronary arteries.
- Characteristic TOE findings Palmon et al list investigations for gas embolism, in order of their sensitivity. TOE turns out to be the best, demonstrating classical echogenic bubbles. This technique can detect as little as 0.02ml/kg of gas. The added bonus is being able to assess for paradoxical embolism.
Management of gas embolism
The following strategies are anaesthetic-sounding because they were written by an anaesthetist to help manage a complication during anaesthesia. However, they are worth knowing about for the ICU environment.
- Disconnect the nitrous oxide. This gas is notorious for collecting in air-filled cavities. N2O will rapidly diffuse into trapped air bubbles and increase the size of the embolus.
- Put the patient in a flat (supine) position. A head-down position was once recommended, for two main reasons. Firstly, any air bubbles in the arterial circulation should spare the brain and bubble up into the gut and legs. Secondly, the gas should collect in the RV apex, where it will hopefully stay - the base of the RV will still be full of blood and the RV will still have something to pump.
In actual fact, this turns out to be complete bullshit. The buoyancy of gas bubbles is not sufficient to counteract blood flow propelling such bubbles toward the head, so you won't save the brain this way- but you certainly will exacerbate cerebral oedema.
- Aspirate the gas using a right atrial catheter (or a long CVC, or a PA catheter).
- Crank the FiO2 to 100%. Apart from being a stereotyped knee-jerk response to hypoxia, this manoeuvre helps to decrease the volume of the gas bubbles, provided atmospheric air is the embolic source. The administration of oxygen as the sole respiratory gas will lead to an increased gradient for nitrogen to exit the air bubbles. An excellent article by Van Liew et al (1992) discusses this phenomenon in luxurious detail.
- Hyperbaric oxygen therapy may be useful, particularly for paradoxic cerebral air embolism. The aim is to decrease the size of the bubbles by creating a nitrogen gradient (as above), but also to improve the oxygenation of ischaemic tissue (i.e. the brain). R.E. Moon (2014) discusses the use of hyperbaric oxygen in this setting; apparently the otcome is better with this "recompression" therapy. apparently, it is indicated even after significant delay, and even when there no apparent air in the CT of the brain.
- Antiepileptic therapy - specific for cerebral gas embolism; these people are prone to seizures.
- Anticoagulation: the air-blood interface is particularly thrombogenic, and some have recommended heparin infusion. As far as I can tell, that recommendation comes on the basis of a single rabbit study (Ryu et al, 1996).
- Corticosteroids: There had at one stage been a belief that steroids may control cytotoxic cerebral oedema. Both the evidence for efficacy and the theoretic rationale are fairly flimsy.
- Lignocaine is though to be beneficial as a "neuroprotective" agent, and was quite popular in the 1980s. Again, mainly animal studies (eg Evans et al, 1984). The NEJM review concludes that "a strong argument can be made" in support of this therapy.
Aeromedical retrieval of the patient with gas embolism
This finally answers Question 21 from the first paper of 2017, in which a cerebrally embolised SCUBA diver is waiting for you to organise their retrieval to the nearby hyperbaric chamber (300km away).
- Hypobaric conditions of flight: air bubbles will enlarge.
- Isobaric flight is dangerous; to put the patient in a helicopter and then fly at low altitude risks the lives of the retrieval crew.
- Transfer by aircraft with a sea-level pressurised cabin is the ideal. This has many problems, for instance 30% extra fuel use due to a lower maximum altitude ceiling. This has implications for retrieval range.
- Vibration: in-flight turbulence leads to increased tribonucleation, whereby gas bubbles precipitate out of a solution. Think of the gas embolism patient being shaken up like a can of Coke.
- Exacerbation of delirium: The neuropsychological changes found in some DCI victims may be compounded by the aeromedical transfer.