High-Flow Nasal Oxygen versus Mechanical Ventilation Through a Laryngeal Mask During General Anesthesia Without Muscle Paralysis: A Randomized Clinical Trial

Authors: Luciano Frassanito, Domenico Luca Grieco, Francesco Vassalli, Alessandra Piersanti, Marco Scorzoni, Francesca Ciano, Bruno Antonio Zanfini, Stefano Catarci, Ursula Catena, Giovanni Scambia, Massimo Antonelli, Gaetano Draisci

Abstract

BACKGROUND:

Apneic oxygenation with high-flow nasal oxygen is a novel intraoperative respiratory support strategy for patients undergoing general anesthesia, but data about its clinical effects are scarce. We conducted a randomized trial to assess whether high-flow nasal oxygen is noninferior to mechanical ventilation through a laryngeal mask in terms of success rate of intraoperative respiratory support among patients undergoing a 30-minute general anesthesia session.

METHODS:

Single-center, randomized, noninferiority trial conducted in Italy between May 2022 and June 2023 and involving American Society of Anesthesiologists class I and II patients undergoing general anesthesia for operative hysteroscopy. Participants were randomized to receive laryngeal mask ventilation (volume-controlled ventilation to obtain end-tidal carbon dioxide between 35 and 45 mm Hg, inhaled oxygen fraction to achieve peripheral oxygen saturation greater than 95%) or high-flow nasal oxygen (70 L per minute, inhaled oxygen fraction of 100%) for intraoperative respiratory support. Patients received general anesthesia with propofol target-controlled infusion without neuromuscular blockade. Primary outcome was intraoperative respiratory support success rate, which was defined as peripheral oxygen saturation greater than 94% and transcutaneous carbon dioxide lower than 65 mm Hg with no need for rescue airway interventions for the entire procedure. Secondary outcomes included the rate of airway-related complications (including need for bag-mask or laryngeal mask ventilation, or tracheal intubation), postoperative respiratory symptoms, and postoperative dyspnea.

RESULTS:

All 180 patients who were randomized completed the trial (90 patients in each group). Median [interquartile range] anesthesia duration was 25 [20–36] minutes in high-flow group and 32 minutes [27–44] in the laryngeal mask group. Intraoperative respiratory support was successful in 89 patients (99%) in both groups (absolute difference 0, unilateral 95% confidence interval, 3%, noninferiority P < .001). Incidence of postoperative respiratory symptoms was significantly lower in high-flow versus laryngeal mask group (2% vs 19%, P < .001), while airway-related complications and postoperative dyspnea were not different. Intraoperative transcutaneous carbon dioxide was significantly higher in high-flow group, with 43% of patients showing values greater than 55 mm Hg.

CONCLUSIONS:

High-flow nasal oxygen is noninferior to laryngeal mask ventilation for intraoperative respiratory support during 30-minute general anesthesia without muscle paralysis. The risk of hypercarbia warrants careful patient selection and monitoring.

BACKGROUND

Use of high-flow nasal oxygen (HFNO), which involves the delivery of heated and humidified oxygen-enriched air at flow rates between 40 and 70 L/min through dedicated nasal cannula, is common in critically ill patients with high respiratory demand.^1^ Recently, this technique has found application in the field of anesthesiology.^2^

Evidence supporting the efficacy of HFNO in the perioperative setting has expanded, showing promising outcomes in various clinical settings, including difficult airway management, procedural sedation for endoscopic procedures, cardiac device implantation, sedated hysteroscopy for assisted reproduction, and brief laryngeal surgery under general anesthesia.^2^

Operative hysteroscopy, a common procedure for investigating abnormal uterine bleeding, performing biopsies, treating endometrial lesions, or evaluating uterine reproductive capacities, is typically conducted in a day-surgery regimen under general anesthesia with ventilation facilitated by face or laryngeal masks.^3^ While laryngeal masks offer effective airway management and allow for hands-free operation by the anesthesiologist, challenges such as improper placement and associated complications do exist.^4^ In a pilot study, we demonstrated that HFNO as sole respiratory support technique may be a safe and feasible alternative for oxygenation and CO2 washout during a 30-minute general anesthesia session without muscle paralysis^5^; potential benefits of this approach would be lower invasiveness and simpler bedside use. However, no study ever assessed whether intraoperative respiratory support with HFNO may be noninferior to conventional mechanical ventilation through laryngeal mask during short-duration general anesthesia in terms of patient-centered clinically relevant outcomes.

We conducted a randomized trial to assess whether HFNO is noninferior to mechanical ventilation through a laryngeal mask in terms of overall success of intraoperative respiratory support during hysteroscopic surgery.

METHODS

Study Oversight

This single-center randomized trial was conducted in a university hospital in Rome, Italy, between September 2022 and June 2023. This study was approved by the University’s Institutional Review Board (ID 4872, protocol number 15190/22), and written informed consent was obtained from all subjects participating in the trial. The trial was registered before patient enrollment at clinicaltrials.gov (NCT05550584), Principal Luciano Frassanito, Date of September 20, 2022). The study was conducted following the guidelines of the Declaration of Helsinki.

Patients

American Society of Anesthesiologists physical status I and II adult patients aged <70 years old, scheduled for elective operative hysteroscopy, were considered for enrollment. Exclusion criteria were New York Heart Association class > II, chronic obstructive pulmonary disease (COPD) with long-term oxygen therapy, obstructive sleep apnea syndrome (OSAS), body mass index (BMI) >30, pregnancy, preexisting cardiac arrhythmias, chronic renal failure, neuromuscular diseases, anticipated prolonged (≥60 minutes) or complex hysteroscopies, upper airways disease.

Procedures

On arrival in the operating room, patients were positioned in the lithotomic position. A peripheral venous cannula was inserted into the hand or forearm, and infusion of 500 ml Ringer Lactate solution commenced. Preoperatively, patients received Omeprazole 40 mg and Dexamethasone 4 mg.

Patients were then randomized in 1 ratio to the HFNO or to the laryngeal mask ventilation group.

Randomization was performed through a computer-based randomization sequence (Randomization.com) kept in sequentially numbered, opaque, sealed and stapled envelopes. Due to the characteristics of the devices and of the ventilation techniques, blinding was not possible.

In the HFNO group, preoxygenation was achieved using 100% oxygen delivered at 30 L/min with dedicated nasal cannula (THRIVE-Fisher and Paykel Healthcare) for 3 minutes. Subsequently, on induction of general anesthesia, oxygen flow was increased to 70 L/min and maintained throughout the procedure. Humidification chamber temperature was set at 31°C, 34 °C, or 37°C based on patient’s comfort.

In the event of oxygen saturation (Spo2< 94%), transcutaneous carbon dioxide (tcCO2)>65 mm Hg, or hemodynamic instability, the attending anesthesiologist provided appropriate pharmacological or rescue interventions. In such cases, bag-and-mask ventilation or endotracheal intubation mechanical ventilation was allowed according to the preference of the attending anesthesiologist.

In the laryngeal mask group, preoxygenation was conducted for 3 minutes via a facial mask delivering 100% oxygen at 15 L/min. After induction of anesthesia, patients were mechanically ventilated through an I-Gel laryngeal mask connected to a mechanical ventilator, which was set in volume-controlled mode. Settings were as tidal volume of 8 ml/kg of predicted body weight; inspired oxygen fraction to maintain Spo2≥96%; positive end-expiratory pressure of 5 cmH2O; respiratory rate adjusted to maintain end-tidal CO2 between 35 and 45 mm Hg. In case of desaturation (defined as Spo2<94%) despite FiO2>80%, tracheal intubation was performed.

In both groups, general anesthesia was standardized. General anesthesia was induced using target-controlled infusion of propofol at an effect site concentration of 7 µg/ml through the Orchestra Infusion system (Fresenius Kabi, Bad Homburg, Germany), along with a bolus of fentanyl at 1 µg/kg. Anesthesia maintenance was obtained with target-controlled infusion of propofol at an effect site concentration of 4 µg/ml. Propofol infusion ceased on completion of surgery or when the patient met awakening criteria.

Postoperatively, patients received Acetaminophen 1g, Ondansetron 4 mg, and Ketorolac 30 mg as part of routine clinical practice. Patients were transferred to the Post Anesthesia Care Unit at the end of the procedure, and discharged after 3 hours of postoperative monitoring if no complications occurred.

Standard intraoperative monitoring included noninvasive blood pressure measurement, 3-lead electrocardiogram, pulse oximetry, and bispectral index monitoring. Spo2, heart rate and tcCO2 was continuously monitored through the Tosca TCM5 monitor (Radiometer, Germany) in both study groups. After appropriate placement of the sensor on the earlobe, tcCO2 was measured at 1-second intervals. Data on Spo2, tcCO2 and heart rate were recorded and stored on a personal computer for offline review.

Data on patient baseline characteristics, comorbidities, surgery indications, laryngeal mask positioning, anesthesia duration, surgical duration, and postoperative hospital stay were collected. Oxygenation, ventilatory, and hemodynamic parameters were recorded during surgery.

Outcomes

The primary outcome measure of the trial was intraoperative respiratory support success rate. Success was defined as Spo2 ≥ 94%, tcCO2<65 mm Hg, and no need for rescue airway interventions during the entire procedure. An assessor blinded to study group assignment reviewed a posteriori recorded data on Spo2, heart rate, and tcCO2 to determine whether the patient had met the criteria to define the primary outcome.

Secondary outcomes included airway-related complications (including any of the need for bag-mask ventilation, laryngeal mask ventilation or tracheal intubation during the procedure), postoperative respiratory symptoms (including any of the cough, sore throat, dysphagia, dysphonia, laryngospasm, oxygen desaturation defined as Spo2<94% after the end of surgery) and postoperative dyspnea (Borg Dyspnea Score>0). As safety end points, we also evaluated Spo2 and tcCO2 values over the course of the study, the duration of surgery and duration of anesthesia in the 2 study groups. Secondary outcomes were evaluated in a standard manner, assuming no difference between groups as null hypothesis and bilateral confidence intervals (CIs). All other outcomes should be considered as exploratory. No adjustment for multiple comparisons was performed.

Sample Size Calculation and Statistical Analysis

Sample size calculation was based on the results of a preliminary feasibility pilot study.^5^ Based on these data, we estimated a 95% success rate both for HFNO and laryngeal mask ventilation. For a unilateral 95% CI and a power of 90%, 82 was the minimal number of patients per group to demonstrate noninferiority, whose limit of clinical relevance was assumed to be 10%. The sample was increased to 90 patients per group to take into account a 10% dropout rate.

Data are displayed as mean ± standard deviation or median (interquartile range) for numerical data or number of events (%) for categorical data. The normality distribution of numerical data was assessed with Shapiro–Wilk test and visually by histograms.

The balance of baseline characteristics between the 2 randomization groups was evaluated with absolute standardized differences (ASD), using means for continuous data and proportions for categorical data. Variables with ASD higher than 1.96·√(2/n) = 0.292 were considered imbalanced (n = 90, group sample size).

The Farrington-Manning test was used for the primary outcome analysis. It estimates a Z-statistic for rate differences adjusted for the noninferiority margin (δ = 0.1): the null hypothesis was that the true rate difference of control minus HFNO group was ≥0.1, while the alternative hypothesis (alternative = “less”) that this value is smaller than the prespecified margin.^6^ Continuous variables were compared with Student T-test or Mann Whitney test, as appropriate; equality of variance was evaluated with the variance ratio test; categorical variables were evaluated with the χ^2^ test or Fisher exact test in case of expected frequencies<5. Intergroup differences are displayed as absolute risk differences with 95% CIs for categorical variables while continuous, not normally distributed variables were compared using 2-sample Wilcoxon rank-sum test and Hodges–Lehman estimation of location shift with corresponding asymptotic 95% CI.

Continuous data regarding tcCO2 and Spo2 were analyzed with ANOVA for repeated mean (±standard deviation) are displayed along with intergroup differences and 95% CIs.

Data were analyzed according to an intention-to-treat principle.

P-values < .05 were considered statistically significant. Data analysis was performed using R (R Foundation for Statistical Computing; version 4.1.2, packages Rcmdr, dplyr, tidyr, ggplot2, effectsize, stddiff, fmsb, DescrTab2).

RESULTS

Between May 2022 and June 2023, 241 patients were assessed for of these, 180 were consented and randomized to the HFNO or laryngeal mask group (Figure 1). All randomized patients completed the study.

Participants’ demographics and clinical characteristics are displayed in Table 1: the 2 cohorts were well balanced except for BMI, slightly higher in the LMA group compared to the HFNO group [24 (4) kg/m^2^ vs 23 (3) kg/m^2^, ASD 0.322].

All patients in the HFNO group received pure oxygen at a flow rate of 70 L per minute. In patients in the laryngeal mask group, median [interquartile range] tidal volume and respiratory rate were 400 mL [375–425] and 13 breaths per minute [12–14].

Study outcomes are reported in Table 2.

Primary Outcome

The success rate was 0.99 (89/90 patients) for both groups with absolute difference 0; the upper limit of the unilateral 95% CI was 0.038. For HFNO, this was less than the assigned noninferiority margin of 0.1 with p for noninferiority 0.001 (Figure 2).

One patient in the HFNO reached tcCO2>65 mm Hg, and bag-mask ventilation was successfully applied to revert hypercarbia. One patient in the laryngeal mask group required intubation due to severe laryngospasm and desaturation (minimum Spo2 87%), reasonably because of fluid overload to expand the uterine cavity.

Secondary Outcomes

The number of patients that experienced postoperative respiratory symptoms in the postanesthesia care unit arrival was 2 (2%) in the HFNO group vs 19 (21%) in the laryngeal mask group, a difference that was statistically significant (absolute difference −19 % [95% CI, −28 to −10], P < .001), with a mild sore throat being the most frequently reported symptom by patients in the latter group. The incidence of postoperative dyspnea and the need for O2 supplementation for transient desaturation was not different between groups.

Exploratory Outcomes

Duration of surgery was comparable between the 2 groups; duration of anesthesia was 25 minutes [20–36] in HFNO group and 32 [27–44] minutes in laryngeal mask group, a difference that was statistically significant (mean difference −7 minutes [95% CI, −10 to −4]); patients in HFNO group had also shorter time from induction of anesthesia to insertion of hysteroscope and shorter time to end of surgery to arousal.

The time course of Spo2 and tcCO2 absolute values is shown in the Supplemental Digital Content, Supplemental Figures 1–2, https://links.lww.com/AA/F365. Average intraoperative Spo2 was slightly but significantly higher in HFNO versus laryngeal mask 99.8% (±0.6%) vs 99.4% (±0.6%), mean difference 0.4% [95% CI, 0.2–0.6], ANOVA P < .001 (Figure 3). While tcCO2 was remarkably stable in the laryngeal mask group during the procedure, it increased steadily in the first 5 minutes after induction of anesthesia in the HFNO group and then plateaued. Overall, average intraoperative tcCO2 was significantly higher in HFNO versus laryngeal mask 42 mm Hg (±5 mm Hg) vs 36 mm Hg (±5 mm Hg), mean difference 6 mm Hg [95% CI, 4–8], ANOVA P < .001 (Figure 3). Median peak intraoperative tcCO2 during anesthesia in the HFNO group was 51 (47–56) mm Hg: overall, 32% of the patients reached tcCO2 in the 55–60 mm Hg range and 11% reached tcCO2 in the 60 to 65 mm Hg range for. After the end of surgery, tcCO2 rapidly decreased but, at the end of anesthesia, it was still higher in HFNO group (Table 2 and Figure 3).

DISCUSSION

In this randomized trial, HFNO was not inferior to laryngeal mask ventilation in enhancing successful respiratory management of patients undergoing operative hysteroscopy under general anesthesia without neuromuscular blockade. The requirement for rescue interventions during the intraoperative period was comparable to that of patients receiving mechanical ventilation through a laryngeal mask. However, average transcutaneous CO2 was higher in patients receiving HFNO, and a significant proportion of patients developed significant accordingly, careful patient selection and monitoring is needed during the treatment.

Hysteroscopic surgery is often performed under general anesthesia, with laryngeal mask ventilation being a common method to ensure airway patency.^7^ However, despite its ease of use, laryngeal mask ventilation can pose risks such as incorrect positioning, which may lead to air leaks and potential hypoventilation. Other risks include partial airway obstruction, aspiration, pneumonia, laryngeal trauma, and injuries to the tongue, nerves, or vocal cords.^8^

In this context, HFNO may offer an attractive alternative, allowing the administration of 40 to 70 L/min of fully heated and humidified air/oxygen mixture at a predetermined FiO2 through specifically designed nasal prongs.^2,8–12^ Compared to conventional oxygen therapy, HFNO limits oxygen desaturation, need for airway interventions, and procedure interruption in patients undergoing surgery under sedation without tracheal intubation.^13,14^ Accordingly, this approach has been proposed to extend safe apnea time during periods of muscular inactivity, demonstrating a delayed onset of hypoxemia and reduced CO2 retention compared to traditional apneic oxygenation techniques.^11,15–17^ Despite several studies investigated CO2 dynamics during apneic oxygenation,^13,18,19^ the precise mechanisms governing gas exchange under these conditions remain not fully understood.^12,20^ Nevertheless, hypercarbia represents the most relevant drawback of HFNO use during general anesthesia, and its occurrence is proportional to the time of apnea and independent from applied flows.^18^ Conversely, the possibility to generate adequate alveolar ventilation with CO2 washout is the greatest advantage of laryngeal mask ventilation. Few clinical trials compared HFNO with laryngeal mask ventilation during general anesthesia. Moreover, while several studies assessed the efficacy of HFNO in terms of intraoperative oxygenation and CO2 washout, none of them systematically investigated HFNO efficacy on more relevant patient-centered clinical outcomes, as the overall success of intraoperative respiratory support or the need for rescue interventions because of inadequate oxygenation/ventilation.^21,22^

Our study revealed that only 1 patient per group required rescue interventions due to these intraoperative complications. In the HFNO group, this was due to the occurrence of hypercarbia; in the laryngeal mask ventilation group, this occurred due to arterial desaturation requiring tracheal intubation. Both patients were successfully treated without postoperative complications. Importantly, the incidence of postoperative respiratory symptoms (including cough, sore throat, dysphagia, dysphonia, laryngospasm or oxygen desaturation) was significantly lower in the HFNO group, suggesting enhanced comfort. Moreover, duration of anesthesia (approximately 30 minutes) was significantly lower in the HFNO group, mostly due to shortened induction and arousal times linked to the simpler respiratory support technique used. Overall, these results indicate that, in patients undergoing general anesthesia without muscle paralysis, HFNO use has an acceptable safety profile, with potential benefits on patient-related clinical outcomes.

Our results are consistent with HFNO physiological mechanisms of action, which include efficient oxygenation and some CO2 washout.^11^ HFNO optimizes apneic oxygenation because the high flows optimize alveolar partial pressure of O2, which remains higher than that in the this facilitates continuous oxygen transfer. This is consistent with the alveolar gas equation, which implies adequate oxygenation in case of a continuous flow of O2 in patent airways, even without alveolar ventilation.^15,23–29^ Moreover, the high flows generate some expiratory pressure preventing or counterbalancing atelectasis,^23^ which otherwise yield interindividually variable lung volume loss.^25,30^ Accordingly, strong evidence supports HFNO as a technique to prolong apnea time with adequate oxygenation. Conversely, progressive development of respiratory acidosis due to CO2 retention, which occurs at an approximative rate of 3 to 5 mm Hg per minute, is the factor that mostly curtails HFNO effectiveness.^18–20,31^ In our study, average tcCO2 values were significantly higher in the HFNO group, while no clinically meaningful differences were detected in Spo2 values. This is consistent with other investigations on the topic.^32,33^ In our cohort, 1 patient in HFNO group required rescue interventions to treat clinically relevant hypercarbia, but a relevant proportion of patients developed tcCO2>55 mm Hg. This warrants accurate patient selection and monitoring.

Overall, however, our findings indicate a median tcCO2 peak value of 51 mm Hg in the HFNO group; HFNO generates CO2 washout due to continuous-flow induced dead space clearance and some alveolar ventilation favored by the interaction between turbulent supraglottic flow vortices and cardiogenic oscillations.^19,34^ Notably, after a progressive rise in tcCO2 values during the first 10 minutes of surgery, these values then stabilized for the rest of the procedure. The most likely explanation for this particular CO2 behavior lies in the possible resumption of some spontaneous ventilatory activity after a threshold of hypercarbia was achieved.^35^ This hypothesis is supported by the physiology of the interplay between hypnotic agents and respiratory drive in the absence of muscle hypnotic agents do not necessarily abolish spontaneous breathing; rather, they reduce the responsiveness of neural centers to hypercarbia. The respiratory drive is then triggered when the pH in the cerebrospinal fluid falls to a lower level.^36–38^ The understanding of the exact mechanism behind this phenomenon warrants further confirmatory investigations. Overall, however, 43% of patients receiving HFNO showed tcCO2 values greater than 55 mm Hg. Further research is also needed to establish what patients are at higher risk of developing hypercarbia during the treatment, which would allow more proper patient selection and treatment individualization.

Strengths of our study include the use of patient-centered outcomes to establish noninferiority, and the blinded evaluation a posteriori of Spo2 and tcCO2 tracings to mitigate the influence of the open-label design on primary end point assessment. Additionally, the standardized methods for administering general anesthesia and conducting study procedures enhance both reproducibility and external validity, particularly for general anesthesia for other surgical procedures having similar duration.

Our study has first, the absence of blood gas analysis may have underestimated the increase in Paco2 in HFNO group; however, continuous assessment of Paco2 is not always feasible in clinical practice, especially in a day-case mini-invasive surgery setting. A method to estimate CO2 levels during the intraoperative use of HFNO is end-tidal CO2 levels. However, this approach is prone to significant error and may underestimate the actual CO2 accumulation rate by as much as 50%.^39^ Differently, tcCO2 offers a reliable and noninvasive alternative, providing accurate Paco2 estimation.^40^ Second, the single-center design, which may restrict the generalizability of our however, the methodology followed strict and standardized procedures that are fully detailed in this article; this should facilitate reproducibility of the findings also in other centers. Third, all patients were female; however, we deem there is no reason to hypothesize that male patients would follow different respiratory physiology, and hence behave differently. Fourth, accurately identifying the eventual resumption of spontaneous ventilatory activity in the HFNO group was not possible. Fifth, the exclusion of patients with obesity, COPD, or known or suspected OSAS—conditions associated with a higher risk of airway closure after anesthesia and hypoventilation when exposed to elevated inspired oxygen fractions—limits the generalizability of our findings. These risks stem from mechanisms such as impaired hypoxic pulmonary vasoconstriction, the Haldane effect, absorption atelectasis, and worsened ventilation-perfusion mismatch.^41,42^ Further research is needed to evaluate whether our results are applicable to this patient population. Sixth, noninferiority was established based on clinically acceptable tcCO2 threshold of 65 mm Hg, which is liberal and somehow arbitrary; the acceptability of this choice was supported by the absence of complications or CO2 retention in any patient after surgery, reinforcing the validity of this definition. Finally, the use of HFNO entails the consumption of 70 L of pure oxygen per minute, leading to significant environmental and economic costs. Whether these costs are acceptable remains to be established, particularly in the absence of clear benefits on robust patient-related outcomes when compared to a safe and effective alternative such as the laryngeal mask. Properly designed studies are warranted to investigate HFNO overall cost-effectiveness in the context of general anesthesia.

CONCLUSIONS

During general anesthesia without muscle paralysis, HFNO proved to be noninferior to laryngeal mask ventilation in terms of intraoperative respiratory support success, with similar needs for rescue interventions.

DISCLOSURES

Conflicts of Interest: L. Frassanito received personal fees by Edwards Lifescience. D. L. Grieco has received payments for travel expenses by Getinge, Fisher and Paykel and Hamilton, personal fees by GE, Intersurgical, Fisher and Paykel, and MSD. M. Antonelli has received payments for Board participation from Maquet, Air Liquide, and Chiesi. G. Draisci received a research grant by Edwards Lifescience and by Biogen outside the submitted work, and a research grant by Fisher and Paykel Healthcare for the conduction of the submitted study. D. L. Grieco and M. Antonelli disclose a research grant by General Electric Healthcare. No other authors declared Conflicts of Interest. Funding: The study was supported by an unrestricted research grant by Fisher and Paykel Healthcare, who had no role in study design, data analysis, article drafting nor in the decision to submit the study for publication. **This manuscript was handled ** Narasimhan Jagannathan, MD, MBA.

Supplementary Material