In this study, we used propofol infusion to maintain deep sedation during endovascular surgery. Relaxation of the tongue and pharyngeal muscles causes narrowing or closing of the upper airway space, resulting in upper airway obstruction in patients under deep sedation. Although a simple airway maneuver, such as head tilt, chin lift, or jaw thrust, may be effective to relieve this obstruction, it is often inconvenient to perform during the surgical period. In our study, the NHF system helped reduce the need to open airway obstruction and reduce the incidence of desaturation compared with a conventional nasal cannula. Improving gas exchange could be achieved by an NHF cannula-induced reduction in rebreathed CO2 volume by clearance of dead space [21, 22] and an increase in end-expiratory lung volume .
NHF can also generate PAP. Many studies on NHF cannulas have shown that this device could generate mean pressure in the pharyngeal space up to 7.1 cmH2O when flow was delivered up to 50 L/minute [11,12,13, 24]. In this study, the NHF group were administered at a flow rate of 60 L/minute that higher than the previous studies [11,12,13, 24]. The higher rate might be helpful in relieving upper airway obstruction and reduce the requirement for performing airway maneuvers in deeply sedated patients in NHF group. The Starling resistor model of the upper airway is used to explain the collapsible (pharyngeal) segment between two rigid tubes (nasal and trachea) . If the airway pressure in upstream is more than critical closing pressure (pressure required to collapse the airway), the obstruction will be relieved and airflow can pass through the lower airways. The study by Lin Y et al.  in outpatients undergoing routine gastroscopy with propofol sedation that used the nasal high flow (NHF) therapy at the flow rate up to 60 L/minute showed the incidence of hypoxia of 0% in NHF group while the desaturation rate in NHF group in the present study was 27.78%. The higher desaturation rate might be from the differences in oxygen concentration (FiO2) and patient positioning. FiO2 in our study was 0.28 and all patients were in supine position, whereas the study by Lin et al. applied FiO2 of 1.0 and all patients were in lateral position. In addition, the study patient characteristics were also different. Almost all patients in our study were ASA physical status grade 3 while the study by Lin Y et al. included the patients with ASA physical status grade 1 to 2 only.
A previous study  showed that an upstream pressure equal to 11.8 ± 2.7 cmH2O was sufficient to maintain airway patency. Mathru M et al. found that nasal continuous PAP (CPAP) can restore the patency of the pharyngeal airway in patients sedated with propofol . In our study, there were 6/18 patients in the NHF group required an airway maneuver. A reason of airway maneuver requirements could be from insufficient PAP level generated by NHF to open the collapsed airway, which might require upstream pressure of more than 12 cmH2O, as mentioned above. The PAP created by NHF depends on whether the person is breathing with the mouth opened or closed and the flow rate used.
In this study, there was no control of the mouth position. Additionally, because sedated patients might have had some mechanical obstruction at the nasopharynx, base of the tongue, and hypopharynx, the NHF could not combat the resistance from those obstructions. Therefore, the airflow proximally leaked from the airway obstruction area. Subsequently, the NHF system was unable to maintain PAP all of the time. In the present study, although the incidence of desaturation and upper airway obstruction in NHF group was significantly lower than NC2 group, however, some patients in NHF group required combination of the interventions. Of the 5 patients having desaturation in NHF group, 3 patients had airflow leak and required both the airway maneuver and airway instrument insertion to maintain SpO2 over 92%. Two of the 3 patients required nasopharyngeal airway insertion only and one patient needed both nasopharyngeal and oropharyngeal airway insertion. In addition, one patient in NHF group experienced apnea during the deep sedation. Recent studies demonstrated that transnasal humidified rapid insufflation ventilatory exchange (THRIVE), an oxygenation technique that delivers continuous, warm and humidified oxygen at a high flow rate (up to 120 L min−1)  via high flow nasal cannula could prolong the safe apnea time both adults  and children . The THRIVE technique is easily implemented method to achieve oxygenation and ventilation without an invasive device . Further research into THRIVE by comparing with current oxygenation technique with NHF cannula is required to demonstrate the benefit of THRIVE in the patients undergoing surgery under deep sedation.
In this study, post-intervention arterial blood gas analysis showed no significant difference in partial pressure of oxygen (PaO2) and carbondioxide (PaCO2) between using the NHF system and a conventional nasal cannula. Although PaO2 and PaCO2 were not correlated with clinical desaturation and the need to perform airway interventions, which had a significantly higher incidence in the NC2 group than in the NHF group. The reason for this finding could be because there were patients with severe hypoxemia in the NC2 group. Therefore, we had to prevent further desaturation by performing an airway maneuver and inserting an airway device before collecting blood samples. For this reason, an accurate measurement of oxygenation and ventilation could not be obtained at this time.
In the NC2 group, two patients had an extremely high PaO2 and PF ratio, which caused an increase in overall oxygenation. As a result, oxygenation in the NC2 group was not significantly different compared with that in the NHF group. Generally, a conventional nasal cannula should only be used with an oxygen flow rate of 6L/minute because exceeding 6 L/minute causes dryness of the nasal mucosa. A previous study reported that an oxygen flow rate of up to 50 L/minute could be delivered by a conventional nasal cannula, but only when the gas was optimally warmed and humidified . In our study, only one patient complained about airway dryness, and it did not lead to low satisfaction by the patient while he was applied oxygen via an NHF cannula.
A previous randomized controlled study showed benefit and safety of the NHF oxygen therapy in hypoxemic patients after receiving cardiothoracic surgery  and a prospective cohort studies shown that NHF cannula was a useful tool as an adjuvant or main oxygen therapy during induction of general anesthesia, maintenance of deep intraoperative sedation, and during early postoperative care . However, our study would be the first randomized controlled study comparing nasal high flow cannula versus conventional nasal cannula in the patients undergoing endovascular under IVS. Although the sample size was small, the significant differences in desaturation, upper airway obstruction, and airway maneuver were detected. The study with larger sample size could give more precise estimates of effects (narrower confidence intervals). Despite the positive results, this study has some limitations. We could not properly control the timing of drawing arterial blood gas because of ethical issues about the safety of patients after desaturation occurred. The apnea time and the length of desaturation was not collected and the screening tools for the risk of obstructive sleep apnea (OSA) e.g., STOP-Bang Questionnaire was not applied. Moreover, further studies should measure the nasopharyngeal pressure to clarify if the NHF system was able to maintain PAP or not.
In conclusion, the use of NHF cannula in patients undergoing endovascular surgery under deep sedation reduced desaturation events and required fewer airway interventions than NC2 with no difference in arterial blood gas analyses and mouth dryness.