Carbon dioxide clearance during apnoea with high‐flow nasal oxygen: epiphenomenon or a failure to THRIVE ?

High-flow nasal oxygen administration during apnoea was first described in a case series of 25 adults in Anaesthesia in 2015 1. Subsequently, case series by other groups 2, 3, randomised trials 4-6 and widespread international experience have confirmed beyond doubt that this is a highly effective approach to apnoeic oxygenation. However, the data surrounding the concept of carbon dioxide clearance during apnoea with high-flow nasal oxygen is more limited, raising the possibility that recent modelling 7 and simulation 8 exercises are premature, and that appropriately conducted randomised trials in adults are still required. It is thus timely to re-examine the strengths and weaknesses of the carbon dioxide data during apnoea, in contrast to the more established benefits associated with high-flow nasal oxygen. The extensive apnoeic oxygenation benefits with high-flow nasal oxygen were recently summarised by Lyons and Callaghan in an article highlighting the evidence supporting use during airway management, especially in high-risk groups, and for tubeless anaesthesia 9. Other physiological benefits of high-flow nasal oxygen are apparent in spontaneously-breathing patients, including: reduction of functional dead space; generation of flow-dependent positive end-expiratory pressure; and decreased work of breathing. These diverse spontaneous ventilation benefits may be more accurately described as high-flow nasal therapy, and show promise in improving patient-centred outcomes when applied postoperatively 10. Overall, it is clear that the development of this technology represents a significant advance in a myriad of settings, and further applications are likely to be discovered. In their original case series, Patel et al. measured end-tidal (n = 20) and arterial (n = 4) carbon dioxide levels during surgery performed under apnoea of varying duration 1. The median apnoea time without significant arterial oxygen desaturation was 14 min and ranged from 5 to 65 min. Carbon dioxide at the end of apnoea was plotted against the apnoeic time, and the slope of the linear regression line was used to calculate an average rate of carbon dioxide increase of 0.15 kPa.min−1. This carbon dioxide accumulation rate appeared low in comparison with a selection of trials published in 1959 (0.45 kPa.min−1) 11, 1961 (0.4 kPa.min−1) 12, 1973 (0.43 kPa.min−1) 13, 1989 (0.45 kPa.min−1) 14 and 2013 (0.24 kPa.min−1) 15. Thus, the concept of transnasal humidified rapid-insufflation ventilatory exchange (THRIVE) was born. The carbon dioxide data obtained by Patel et al. have several limitations. First, the majority of carbon dioxide estimates were based on end-tidal levels after placement of a definitive airway. This approach does not account for the increasing arterial-to-alveolar gradient with progressive apnoea and is vulnerable to error 9. Indeed, Gustaffson et al. have since shown in a similar case series that end-tidal measurements underestimate the true arterial carbon dioxide accumulation rate by approximately 50% (0.12 kPa.min−1 vs. 0.24 kPa.min−1), whereas transcutaneous carbon dioxide measurements serve as an excellent and preferred non-invasive surrogate 2. Second, the use of historical controls, rather than contemporaneous controls as in a randomised trial setting, introduces significant confounding. Varying patient populations undergoing anaesthesia with varying drug regimens under varying conditions may well have varying rates of carbon dioxide accumulation during apnoea, irrespective of the apnoeic oxygenation technique used. For example, many of the historical controls cited above used suxamethonium as a bolus and infusion to prevent spontaneous ventilation, and this exposure in itself is known to increase carbon dioxide production 16. Perhaps most importantly, the rate of carbon dioxide increase during apnoea is non-linear 12. Therefore, the varying durations of the apnoeic period being tested in different trials can confound the average carbon dioxide accumulation rate reported. The non-linear rise of carbon dioxide during apnoea was first described by Eger and Severinghaus in 1961 12 and is a finding replicated by multiple other groups 2, 14 including our own 17. This non-linearity leads to a slowing of carbon dioxide accumulation as the apnoeic period progresses and has a significant impact on the 'average' rate of carbon dioxide increase reported for a given period. In our recent evaluation study of buccal oxygenation, we illustrated this by reporting a carbon dioxide accumulation rate of 0.42 kPa.min−1 during the first 4 min of apnoea and 0.29 kPa.min−1 during the complete 12.5 min of apnoea 17. Thus, if a given apnoeic oxygenation device undergoes a short apnoeic test, it will report a relatively higher carbon dioxide accumulation rate than if it undergoes longer apnoeic tests. The average or median apnoeic periods varied considerably between the original case series describing THRIVE (14 min) and the historical control studies used as comparators (Frumin et al., 45 min; Eger et al., 16.4 min; Fraioli et al., 15 min; Stock et al., 3.7 min; and Rudlof et al., 25 min). Clearly, to test the hypothesis that the rate of carbon dioxide accumulation is lower during apnoea with high-flow nasal oxygen, a randomised trial is required where the influence of the apnoeic period is controlled or adjusted for. In contrast to the other clinically available apnoeic oxygenation devices (low flow nasal prongs 18, insufflating laryngoscopes 19, buccal oxygen 20) it is remarkable that high-flow nasal oxygen during apnoea in adults became established into practice on the basis of several case series alone. The first randomised controlled trial in adults with a prescribed apnoeic test has only just been published 6. This confirmed an oxygenation benefit in morbidly obese patients, extending the safe apnoea time from 180 to 261 s, but reported no difference in plateau end-tidal carbon dioxide at the end of apnoea (5.11 kPa vs. 4.99 kPa). The only two randomised trials of high-flow nasal oxygen with arterial or transcutaneous carbon dioxide measurements have been conducted in paediatric populations 4, 5. Humphreys et al. compared high-flow nasal oxygen against a control arm and found no difference in carbon dioxide accumulation rates (0.32 kPa.min−1 vs. 0.32 kPa.min−1). Similarly, Riva et al. in a three-arm trial evaluated high-flow nasal oxygen with 100% oxygen vs. high-flow nasal oxygen with 30% oxygen vs. low-flow nasal prongs with 100% oxygen, and again found no difference in the rate of carbon dioxide increase (0.55 kPa.min−1 vs. 0.60 kPa.min−1 vs. 0.57 kPa.min−1, p = 0.25). The almost two-fold difference in overall carbon dioxide accumulation rates reported in these two trials further demonstrates the point that variations in trial protocols can influence carbon dioxide values greatly, even when the same apnoeic oxygenation technique is used. The lack of impact of high-flow nasal oxygen on carbon dioxide accumulation rates in paediatric populations may be explained by differences in anatomy and physiology that prevent the purported mechanisms of carbon dioxide clearance from applying 21. Equally, these findings are consistent with the concept that high-flow nasal oxygen does not impact on carbon dioxide accumulation in children or adults, and the original observation of 'ventilatory exchange' was an epiphenomenon arising from the limitations discussed above. The hypothesis that high-flow nasal oxygen leads to carbon dioxide clearance from the airway during apnoea in adults and therefore a lower rate of carbon dioxide accumulation can only be tested in an appropriately designed randomised trial. Such a trial must use a robust measure of carbon dioxide (arterial or transcutaneous) and set out to achieve the same apnoeic period in the high-flow nasal oxygen arm and the comparator arm, or to adjust for observed differences. This is perhaps best accomplished by comparing high-flow nasal oxygen to another available apnoeic oxygenation device to avoid premature arterial oxygen desaturation. Using prior studies to estimate data variability, a randomised trial would require a sample size of approximately 20 participants to detect a difference of 0.32 kPa.min−1 vs. 0.24 kPa.min−1 (80% power, p < 0.05). At the time of writing, a search of the trial registries in the USA, Australia and Europe revealed several trials in the active recruitment phase that are appropriately designed and powered to address the carbon dioxide hypothesis (www.clinicaltrials.gov: NCT03478774 22; NCT03550924; both due to complete end-2019). Although the proven oxygenation benefits of high-flow nasal oxygen will continue to be captured and refined by anaesthetists, evidence that can reliably guide practice with respect to ventilatory exchange during apnoea is eagerly awaited. AT has been an investigator on two randomised controlled trials of apnoeic oxygenation with a buccal device in the oral cavity. There are no commercial interests or intellectual property associated with these studies. No external funding or competing interests declared.