Gas composition and pressure in the hypopharynx during high-flow oxygen therapy through a nasal cannula in healthy volunteers with different breathing patterns

Abstract

Background

High-flow nasal cannula is widespread in patients with hypoxemic and hypercapnic respiratory failure, but physiological data concerning influence of the combination of breathing pattern, preset flow rate (PFR), and inspiratory oxygen fraction (F_DO₂) on end-expiratory pressure (EEP), capnogram, oxygram, and exhaled tidal volume (VTe) remains insufficient.

Methods

The study included 20 healthy subjects with 12 combinations of PFR (30–60-80 L/min) and F_DO₂ (40–60-80–100%) multiplied by 4 breathing patterns: mouth closed (CM), mouth open (OM), and combination of the CM and OM with hyperpnea (HCM and HOM). Pressure, capnogram, oxygram were measured from hypopharyngeal catheter, VTe, and subject’s comfort were assessed.

Results

Inspiratory oxygen fraction (FiO₂) were close to F_DO₂ at the PFR of 30 L/min (CM), and 60 L/min (HCM). FiO₂ during the OM and HOM were much less than F_DO₂, variable and unpredictable. PFR of 60 L/min was sufficient to keep FiO₂ close to F_DO₂ during the CM and HCM. End-expiratory carbon dioxide (F_ECO₂) decreased with an increase in the PFR and F_DO₂, reaching 1.4 (1.1–1.7)% at F_DO₂ 100% and PFR of 80 L/min. EEP had grown a lot with the PFR increase and were highly variable reaching 11.1 (7.7–14.8) cmH₂O at the PFR of 80 L/min. VTe at the PFR of 60 and 80 L/min were 948.0 (715.0–1204.8) and 948.0 (869.0–1422.0) ml, respectively. PFR of 60 L/min and 80 L/min were associated with discomfort.

Conclusion

HCM, OM, and HOM in healthy subjects decreased FiO₂ and F_ECO₂ (more pronounced during OM and HOM). HFNC within the CM and HCM provided flow-dependent CPAP-effects over a wide range and could be associated with lung hyperinflation. An excessive PFR led to discomfort.

Trial registration

ClinicalTrials.gov identifier: NCT06189716, registered on 19/12/2023.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12871-025-03267-9.

Background

A high-flow nasal cannula (HFNC) is a promising tool for patients with respiratory failure that uses a combination of a continuously blended flow of air and oxygen delivered through a nasal cannula at preset flow rates close to or higher than the patient’s mean inspiratory flow [1]. Flow from the HFNC device is warmed and humidified according to physiological levels inside the trachea, bronchi, and even alveoli [2]. HFNCs are currently widely used for patients with hypoxemic and hypercapnic acute respiratory failure (ARF), including severe pneumonia, COVID-19, and COPD [3, 4]. The physiological effects of the HFNC are realized through more accurate inspiratory oxygen fraction (FiO₂) titration [5–7], improved carbon dioxide (CO₂) clearance [8–10], an elevated expiratory pressure (EEP) with an increase in the end-expiratory lung volume [11, 12], and better tolerance than noninvasive ventilation (NIV) [13].

Despite these effects and the clinical data concerning the efficacy of HFNCs, real O₂ and CO₂ fractions and end-expiratory pressure (EEP) close to the trachea are not well established, nor are data on the influence of different breathing patterns on these values [3, 4]. Changes in gas composition, expiratory pressure, and inspiratory/expiratory time with changes in breathing patterns could be of utmost importance because of their influences on the clinical efficacy and local or general side effects of HFNC therapy. For example, an increase in a subject’s minute ventilation from normal to 20 L/min during low-flow oxygen therapy through a facial mask (or nasal cannula) may decrease the FiO₂ from 100 to almost 30%, thus greatly reducing the usefulness of oxygen therapy in patients with hypoxemia [14, 15].

Optimal combinations of the preset flow rate and preset oxygen fraction (F_DO₂) in patients with different breathing patterns are urgently needed to improve clinical efficacy and decrease the possible side effects of HFNCs. We performed our study to clarify these important points using simultaneous pressure, respiratory gases close to the trachea (in the hypopharynx), and tidal volume measurements in healthy volunteers with different breathing patterns.

Methods

Study design

The study was approved by the Institutional Ethics Committee (reference number: 25–23, date of approval 14/12/2023) and registered on ClinicalTrials.gov (NCT06189716). Written informed consent was obtained from all healthy volunteers enrolled in the study.

We recruited healthy volunteers (students, residents, and anaesthesiologists from Sechenov University) through oral advertisements and announcements in local social networks. The participants were not compensated. We recruited only nonsmokers or mild smokers (smoking index < 200, where the smoking index = cigarettes per day * years of smoking). Volunteers fasted for no less than 3 h before the study.

Volunteers were seated in a chair. We applied topical anaesthesia with a lidocaine spray in both nostrils, on the surface of the tongue, and the pharynx. After two minutes, a lubricated multiperforated 8 French catheter was inserted into the hypopharynx through a nostril until the tip of the catheter was below the root of the tongue. The appropriate position of the catheter was checked by visual inspection throughout the study (after each step). Before all the measurements, the volunteers remained seated for 5 min to calm their breathing. At baseline, we recorded the atmospheric pressure from the local weather station and measured peripheral oxygen saturation (SpO₂), the inspiratory and expiratory fractions of oxygen (FiO₂ and FeO₂, respectively), the end-tidal carbon dioxide fraction (FeCO₂), the respiratory rate (RR), and the inspiratory and expiratory times (Ti and Te, respectively) using a side-stream multigas analyser (MPR6‒03 monitor, Treaton Electronics, Russia), the expired tidal volume (VTe) using an expiratory flow sensor of the ventilator through a mouthpiece (Savina Select E, Drager, Germany), and the EEP using a pressure transducer (AICU Vision, Bark Technology, Kazakhstan). Data on oximetry, capnography, Ti, and Te were analysed using dedicated software (SideStream Capno, Treaton Electronics, Russia). All measurements were recorded within the stable respiratory pattern viewed on capnography. The catheter was flushed with air on demand. Additionally, we recorded patient comfort using a visual analogue scale (from 1 to 10, where 10 meant full comfort) and asked the subjects to describe their sensations throughout the study period.

After the initial measurements, we established a high-flow oxygen device through a nasal cannula (Hifent HUMID-BH, RespirCare Medical Solutions, China) at a flow rate of 30 L/min and a preset oxygen fraction (F_DO₂) of 40%. Cannula sizes were selected according to the manufacturer’s manual (5 mm diameter for all subjects). We performed 12 series of measurements with preset flow rates (PFR) of 30, 60, and 80 L/min combined with preset F_DO₂ values of 40, 60, 80, and 100% at each flow rate. In each series, we asked the subjects to breathe in four different respiratory patterns: quiet breathing with a closed mouth (CM), quiet breathing with an open mouth (OM), hyperpnea with a closed mouth (HCM), and hyperpnea with an open mouth (HOM). For the hyperpnea series, the subjects were asked to breathe approximately two times deeper (subjects were trained with a volumetric assessment) at a rate of 30 breaths per minute, and the RR in these series was controlled with a multigas monitor. For the open mouth series (OM and HOM), the volunteers were asked to breathe through the mouth. We performed all measurements the same as in the baseline period at each series combined with each respiratory pattern (48 steps total), except VTe during OM and HOM (unable to perform the correct VTe measurements). All the measurements were recorded after 2 min of the current step. The wash-out period between each step was approximately 2 min. The total duration of the measurements for a subject was approximately 4 h. We also assessed oxygram and capnogram waveforms during all steps of the study. Additionally, after completing all the measurements, the volunteers were asked to describe their feelings in their own words. We calculated the modified ventilatory ratio (mVR) at the end of the CM and HCM steps with the following equation: mVR = [RR * VTe * (PetCO₂ (mmHg)—4 mmHg)]/[predicted body weight (kg) * 100 (ml/kg/min) * 37.5 mmHg]. We calculated the mean volunteer flow by dividing VTe by Ti at the end of each step for CM and HCM with the assumption that VTe was close to the inspired tidal volume. The mean subject flow was compared with the preset flow rate and VAS comfort score at all steps.

Statistical analysis

Given the physiological design of the study, we did not perform a formal sample size calculation. We used a sample size of 10 patients for this study, in accordance with previous investigations on this topic [11]. Descriptive statistics included proportions for categorical variables and medians with interquartile ranges (IQRs) for continuous variables. No imputation was performed for missing data. Comparisons between steps were performed using the Friedman test for continuous variables and the Cochran’s Q test for binary values. Post-hoc pairwise comparisons were made using the Friedman test with Bonferroni correction for multiple comparisons (to reduce type I errors). A two-sided p < 0.05 was considered statistically significant. All the statistical analyses were performed using SPSS 26.0 (IBM, Armonk, NY, USA).

Results

We included 20 healthy volunteers (aged 24 (22–36) years, 11 men, height of 176 (167–181) cm, weight of 76.5 (66.0–85.0) kg, body mass index of 25.6 (21.2–28.5) kg/m², two mild smokers). One volunteer refused to continue the study at a PFR of 80 L/min because of severe discomfort but completed all steps with PFR of 30 and 60 L/min (36 steps).

Preset F_DO₂ of 100%

Influences of the PFR and breathing pattern on FiO₂

An increase in the PFR from 30 to 80 L/min during all phases led to a significant increase in FiO₂ (p < 0.05 for all comparisons). During the CM phase, PFR of 60 and 80 L/min provided the preset F_DO₂; at the PFR of 30 L/min, the FiO₂ was close to 100%. An increase in the PFR from 30 to 80 L/min in the OM phase increased the median FiO₂ to 97.8%, but with a wide interquartile range (IQR) (39.4–100.0%), an increase in the PFR from 30 to 60 L/min in the OM phase had a similar effect. During the HCM phase, an increase in the PFR from 30 to 60 and to 80 L/min provided an FiO₂ close to 100%. During the HOM pattern, only an increase from the PFR from 30 to 80 L/min led to a significant increase in FiO₂, but it had a wide range of FiO₂ values and never reached the preset level of F_DO₂. OM and HOM patterns provided less FiO₂ than CM and HCM patterns, respectively, with very wide IQRs (Fig. 1, Table 1).

Fig. 1.

Measured inspiratory oxygen fraction in the hypopharynx at various breathing patterns of volunteers with different preset flow rates and preset inspiratory oxygen fraction of 100%. A CM—quiet breathing with closed mouth. B OM—quiet breathing through open mouth. C HCM—hyperpnea with closed mouth. D HOM—hyperpnea through open mouth. Abbreviations: FiO₂—measured inspiratory oxygen fraction

Table 1.

Gas composition, pressure in the hypopharynx, and respiratory parameters at different preset flows, 100% inspiratory fraction of oxygen, and various breathing patterns of volunteers

Variable	Breathing Pattern (BP)	HFNC flow rate			P between preset flow
		30 L/min	60 L/min	80 L/min
FiO2, %	CM	98.3 (95.3–99.6)	99.9 (98.2–100.0)*	100.0 (99.3–100.0)**	< 0.001
	OM	84.8 (41.7–94.0)†	91.2 (35.6–100.0)*†	97.8 (39.4–100.0)**†	0.021
	HCM	81.4 (77.2–94.3) ‡	96.9 (94.1–99.8)*	99.0 (98.1–100.0)**	< 0.001
	HOM	59.2 (32.8–67.2)#	66.6 (34.1–85.9)#	64.3 (36.3–82.3)**#	0.014
	p between all BP	< 0.001	< 0.001	< 0.001
Abnormal oxygram waveform, n (%)	CM	0	0	0	NA
	OM	7 (35)†	11 (55)†	11 (55)†	0.069
	HCM	13 (65) ‡	15 (75) ‡	16 (80) ‡	0.247
	HOM	15 (75)	17 (85)	17 (85)	0.135
	p between all BP	< 0.001	< 0.001	< 0.001
FeCO2, %	CM	2.9 (2.6–3.6)	2.6 (2.1–3.4)*	2.1 (1.8–3.1)**	< 0.001
	OM	2.6 (2.2–2.9)†	2.2 (1.6–2.9)*†	2.0 (1.7–2.4)**	0.001
	HCM	2.4 (2.2–3.2) ‡	2.0 (1.6–2.6)*‡	1.7 (1.4–2.5) **‡	< 0.001
	HOM	1.9 (1.7–2.2)#	1.8 (1.5–2.3)*	1.4 (1.1–1.7)**§#	< 0.001
	p between all BP	< 0.001	< 0.001	< 0.001
Expiratory pressure, cmH2O	CM	3.8 (3.1–6.6)	7.6 (6.4–11.4)*	11.1 (7.7–14.8)** §	< 0.001
	OM	1.2 (0.7–1.4)†	1.5 (0.9–2.3)†	1.5 (0.8–2.9)†	0.102
	HCM	5.8 (3.5–8.2) ‡	9.4 (6.8–13.1)* ‡	12.7 (8.2–18.1)** § ‡	< 0.001
	HOM	1.5 (1.1–1.7)#	2.4 (1.4–2.7)*#	3.0 (1.3–3.8)**#	0.005
	p between all BP	< 0.001	< 0.001	< 0.001
Delta pressure, cmH2O	CM	3.5 (2.5–5.8)	3.5 (1.9–9.4)	4.5 (1.4–11.4)	0.500
	OM	1.1 (0.8–1.5)†	1.2 (0.7–1.7)†	1.5 (0.6–3.7)†	0.584
	HCM	7.1 (4.5–10.0) ‡	9.9 (6.1–14.7) ‡	12.2 (9.1–18.6) **‡	0.009
	HOM	1.8 (1.3–3.2)#	2.1 (1.6–3.7)#	1.9 (1.1–4.4)#	0.383
	p between all BP	< 0.001	< 0.001	< 0.001
Minimal inspiratory pressure, cmH2O	CM	0.5 (0.3–1.5)	3.9 (2.3–5.1)*	5.4 (1.5–8.0)**	0.001
	OM	0.3 (0.0–0.4)	0.3 (0.0–0.9)†	0.2 (−0.2–1.0)†	0.212
	HCM	−1.9 (−3.9–0.3) ‡	0.6 (−2.5–2.5)*‡	0.5 (−2.4–2.7)**‡	0.025
	HOM	−0.5 (−1.8–0.3)	−0.1 (−1.3–0.5)	0.0 (−0.7–0.5)	0.097
	p between all BP	< 0.001	< 0.001	< 0.001
Exhaled tidal volume, mL	CM	711.0 (608.3–790.0)	948.0 (715.0–1204.8)*	948.0 (869.0–1422.0)**	< 0.001
	HCM	790.0 (681.4–898.6)	987.5 (869.0–1194.9)*	1027.0 (869.0–1264.0)**	< 0.001
	p between BP	0.059	0.808	0.819
Respiratory rate, min−1	CM	11.0 (9.3–16.0)	10.5 (8.0–14.0)	10.0 (7.0–15.0)**	0.043
	OM	14.0 (11.0–16.8)	15.0 (12.0–18.9)	16.0 (12.0–19.0)	0.531
	HCM	27.0 (24.3–30.0) ‡	30.0 (28.0–32.8)* ‡	30.0 (28.0–33.0)** ‡	0.004
	HOM	30.0 (28.0–31.0) #	31.0 (30.0–33.0)*	32.0 (29.0–35.0)**	0.006
	p between BP	< 0.001	< 0.001	< 0.001
Ventilatory ratio (modified), AU	CM	0.62 (0.49–0.91)	0.59 (0.38–0.88)	0.50 (0.40–0.73)	0.076
	HCM	1.13 (0.54–1.54) ‡	1.22 (0.96–1.50) ‡	1.05 (0.71–1.84) ‡	0.949
	p between BP	< 0.001	< 0.001	< 0.001
Expiratory time, s	CM	3.4 (2.2–5.4)	3.4 (2.9–6.0)	3.7 (2.5–5.8)	0.759
	HCM	1.2 (1.0–1.3) ‡	1.0 (0.9–1.2) ‡	1.0 (0.9–1.2) ‡	0.063
	p between all BP	< 0.001	< 0.001	< 0.001
Inspiratory time, s	CM	1.7 (1.2–2.0)	1.7 (1.1–2.0)	1.8 (1.2–1.9)	0.570
	HCM	1.0 (1.0–1.2) ‡	0.9 (0.8–1.0) *‡	0.9 (0.8–1.0)**‡	0.002
	p between all BP	< 0.001	< 0.001	< 0.001
Ti/Te, ratio	CM	0.47 (0.30–0.65)	0.41 (0.34–0.50)	0.49 (0.29–0.60)	0.311
	HCM	0.90 (0.77–1.00) ‡	0.91 (0.83–1.00) ‡	0.83 (0.76–1.00) ‡	0.156
	p between all BP	0.001	< 0.001	0.001
Visual-analog scale of comfort, points	CM	8.0 (7.3–9.8)	7.0 (5.3–8.8)*	6.0 (4.0–8.0)** §	< 0.001
	OM	8.0 (7.0–9.8)	6.5 (5.3–8.8)*	6.0 (3.0–8.0)** §	< 0.001
	HCM	8.0 (6.3–9.0)	6.5 (5.0–8.0)*	6.0 (3.0–8.0)**	< 0.001
	HOM	8.0 (6.3–9.0)	6.5 (5.0–8.0)*	5.5 (3.0–8.0)**	< 0.001
	p between all BP	0.030	0.051	0.866
Mean inspiratory flow, L/min	CM	26.3 (20.3–33.6)	39.6 (27.1–49.8)*	44.8 (30.1–51.8)**	0.003
	HCM	43.6 (40.3–52.1) ‡	61.6 (50.0–77.0)* ‡	68.4 (56.9–90.9)** ‡	< 0.001
	p between all BP	< 0.001	0.001	0.001
Preset flow/mean inspiratory flow	CM	1.14 (0.89–1.48)	1.51 (1.20–2.21)	1.79 (1.54–2.66)** §	0.001
	HCM	0.69 (0.58–0.74) ‡	0.97 (0.78–1.20)* ‡	1.17 (0.88–1.41)** ‡	< 0.001
	p between all BP	< 0.001	0.001	0.001

Data are presented as median (interquartile range)

HFNC high flow oxygen through nasal cannula, BP breathing pattern, CM quiet breathing with closed mouth, OM quiet breathing with opened mouth, HCM hyperpnea with closed mouth, HOM hyperpnea with opened mouth, FiO₂ measured inspiratory oxygen fraction, FeO₂ measured expiratory oxygen fraction, FeCO₂ end-expiratory carbon dioxide fraction, AU arbitrary units, VAS visual-analog scale (10 points—maximal comfort, 1 point—maximal discomfort), Te expiratory time, Ti inspiratory time

Significant differences (Friedman test with Bonferroni correction for multiple comparisons or Cochran’s Q test, where appropriate with p < 0.05) between preset flows with stable respiratory pattern: * 30 vs 60 l/min, ** 30 vs 80 l/min, § 60 vs 80 l/min. Significant differences (Friedman test with Bonferroni correction for multiple comparisons or Cochran’s Q test, where appropriate with p < 0.05) between different respiratory patterns with a constant preset flow: †CM vs OM, ‡CM vs HCM, #HCM vs HOM

Significant differences in p values are highlighted in bold

We discovered a variety of “abnormal” oxygram waveforms during the OM, HCM, and HOM patterns: inverted (almost copying the capnogram shape), flat (without prominent inspiratory and expiratory phases), and combined (combination of inverted and flat) (Fig. 2). The prevalence of abnormal oxygrams was zero during CM breathing and very common during HCM, HOM, and OM breathing (less common than during HCM and HOM breathing) at all PFR (Table 1). An increase in the PFR did not affect the oxygram waveform (Table 1).

Fig. 2.

Capnogram and inverted oxygram during hyperpnea through open mouth. A inverted oxygram pattern. B combined oxygram pattern. Blue line: oxygram, Red line: capnogram

Influences of the PFR and breathing pattern on FeCO₂

The initial FeCO₂ was 4.7 (4.3–5.3)%. The initiation of HFNC therapy resulted in a significant decrease in FeCO₂ (p < 0.001). An increase in the PFR from 30 to 80 L/min resulted in a significant decrease in FeCO₂ during all breathing patterns (p < 0.01 between all comparisons). Compared with the CM pattern, the OM pattern significantly decreased the FeCO₂ level; compared with the CM pattern, the HCM pattern decreased the FeCO₂ level; and compared with the HCM pattern, the HOM pattern decreased the FeCO₂ level (Fig. 3, Table 1).

Fig. 3.

Measured end-expiratory carbon dioxide fraction in the hypopharynx at various breathing patterns of volunteers with different preset flow rates and preset inspiratory oxygen fraction of 100%. A CM—quiet breathing with closed mouth. B OM—quiet breathing through open mouth. C HCM—hyperpnea with closed mouth. D HOM—hyperpnea through open mouth. Abbreviations: F_ECO₂—end-expiratory carbon dioxide fraction

Influences of the PFR and breathing pattern on the end-expiratory pressure and delta pressure

An increase in the PFR to 30, 60 and 80 L/min during the CM pattern resulted in a significant increase in the EEP: 3.8 (3.1–6.6) mbar), 7.6 (6.4–11.4) mbar, and 11.1 (7.7–14.8) mbar, respectively (p < 0.001 for all comparisons). A change in the breathing mode to OM significantly decreased the EEP, and an increase in the PFR did not influence the EEP during OM (Fig. 4A and B, Table 1). A change in the breathing mode from CM to HCM resulted in an increase in the EEP in the hypopharynx (p < 0.05 for all comparisons) (Fig. 4A and B, Table 1). An increase in the PFR during the HCM pattern significantly increased the EEP (p < 0.001 for all comparisons) (Fig. 4A and B, Table 1). The EEP during the HOM pattern was significantly lower than that during the HCM pattern (p < 0.001 for all comparisons). An increase in the PFR significantly increased the EEP during the HOM pattern (p = 0.005 for all comparisons) (Table 1).

Fig. 4.

Measured end-expiratory pressure in the hypopharynx (A, B) and measured expiratory tidal volume (C, D) at various breathing patterns of volunteers and different preset flow rates and preset inspiratory oxygen fraction of 100%. A, C during CM—quiet breathing with closed mouth. B, D during HCM -hyperpnea with closed mouth. Abbreviations: EEP—end-expiratory pressure, VTe—expiratory tidal volume

DeltaP was lower during OM than during CM breathing, higher during HCM than during CM breathing, and lower during HOM than during HCM breathing (Table 1). DeltaP significantly increased with the increasing PFR only during the HCM phase (Table 1).

Influences of the PFR and breathing pattern on the exhaled tidal volume

The initial VTe before the study was 422.7 (347.6–509.6) mL. VTe during CM with an F_DO₂ of 100% increased from the baseline value to 711.0 (608.3–790.0), 948.0 (715.0–1204.8), and 948.0 (869.0–1422.0) mL at PFR of 30, 60, and 80 L/min, respectively (p < 0.001) (Fig. 4C and D, Table 1). An increase in the PFR during the HCM pattern also increased VTe (p < 0.001). The VTe during the HCM and CM patterns did not differ at any of the PFR (p > 0.05) (Fig. 4C and D, Table 1).

Influences of the PFR and breathing pattern on the expiratory and inspiratory times

The initial Te before the study was 2.2 (2.0–2.7 s). The Te during the CM pattern at PFR of 30 L/min increased to 3.4 (2.2–5.4) s (p = 0.005) (Fig. 5A, Table 1). An increase in the PFR to 60 and 80 L/min did not change the Te further during the CM phase (Fig. 5A, Table 1). Compared with the CM mode, switching the breathing mode to HCM resulted in a significant decrease in the Te at all PFR (Fig. 5B, Table 1). An increase in the PFR did not change the Te further during the HCM phase (Fig. 5B, Table 1). The initial Ti before the study was 1.7 (1.1–1.9 s). Ti during the CM pattern and an F_DO₂ of 100% did not change at any of the flow rates (Table 1). A change in the breathing mode from CM to HCM resulted in a significant decrease in Ti at all PFR (Table 1). An increase in the PFR from 30 to 60 and 80 L/min during the HCM pattern shortened the Ti (p = 0.002) (Table 1).

Fig. 5.

A, B Measured expiratory time at various breathing patterns of volunteers and different preset flow rates and preset inspiratory oxygen fraction of 100%. A CM—quiet breathing with closed mouth. B HCM—hyperpnea with closed mouth. C Mean subject’s expiratory flow during quiet breathing with closed mouth at different preset flow rates and preset inspiratory oxygen fraction 100%. D Comfort expressed as visual-analog scale during quiet breathing with closed mouth at different preset flow rates and preset inspiratory oxygen fraction 100%. Abbreviations: VAS—visual-analog scale (10 points—maximal comfort, 1 point—maximal discomfort)

Influences of the PFR and breathing pattern on the respiratory rate

The initial RR was 14 (12–16) min⁻¹. The RR during the CM pattern at a flow rate of 30 L/min decreased to 11.0 (9.3–14.0) min⁻¹ (p = 0.012). An increase in the PFR from 30 to 80 L/min decreased the RR further to 10.0 (7.0–15.0) (p = 0.043) (Table 1). A change in the breathing mode from CM to OM resulted in an increase in the RR at PFR of 60 and 80 L/min, which did not reach statistical significance (p = 0.058 and p = 0.059, respectively) (Table 1). The RR was significantly increased during the HOM breathing pattern compared with that during the HCM breathing pattern at a flow rate of 30 L/min (Table 1).

Influences of the PFR and breathing pattern on the modified ventilatory ratio

The initial mVR was 0.71 (0.58–0.85). The mVR during the CM pattern at all flow rates did not change compared with the initial mVR (Table 1). Switching the breathing pattern from CM to HCM was associated with an increase in the mVR at all preset flow rates (Table 1).

Influences of the PFR and breathing pattern on the subject’s comfort

The VAS comfort score during the CM breathing pattern at a flow rate of 30 L/min was 8.0 (7.3–9.8) points. An increase in the PFR to 60 and 80 L/min during the CM breathing pattern resulted in a significant decrease in the subject’s comfort to 7.0 (5.3–8.8) and 6.0 (4.0–8.0) points, respectively (p < 0.001) (Fig. 5C, Table 1). The same dynamics of the subject’s comfort were observed for all other patterns (Table 1). A change in the breathing pattern from CM to HCM and from OM to HOM did not change the subject’s comfort (Table 1).

The mean subject expiratory flow (EFmean) increased as the PFR increased during the CM and HCM patterns (Table 1). The HCM pattern was associated with a higher EFmean than CM at all PFR (p = 0.001) (Table 1).

The VAS comfort score corresponds to the ratio of the PFR to the EFmean of a volunteer during CM: the maximum comfort of the subject was observed at a ratio close to 1:1, and the maximum discomfort was observed at a ratio of approximately 2:1 (Fig. 5D, Table 1).

The subjects’ answers concerning their discomfort are presented in the Supplemental material (Table S5).

Physiological variables at preset F_DO₂ values of 40%, 60%, and 80%

The dynamics and significant differences of the measured values with the preset F_DO₂ of 40, 60 and 80% were similar to those of the PFR increase and changes in the breathing patterns with the preset F_DO₂ of 100% The FiO₂/F_DO₂ ratio decreased with increasing F_DO₂. For example, an increase in F_DO₂ from 40 to 100% during HOM at PFR of 80 and 60 L/min resulted in a decrease in the median FiO₂/F_DO₂ ratio from 85 to 64% (Tables 1–2, S1–S3; Figs. S1–S2). An increase in the PFR did not influence the measured RR with the preset F_DO₂ of 40, 60 and 80% during all breathing patterns; RR did not differ between the CM and OM or between the HCM and HOM patterns. All data on the preset F_DO₂ values of 40, 60, and 80% are presented in Supplemental Tables S1, S2, and S3, respectively.

Influences of the preset F_DO₂ at constant PFR on FeCO₂, the mVR, and the comfort of the subject

The increase in F_DO₂ during the CM pattern at constant PFR decreased the FeCO₂ and mVR in all comparisons (Supplemental Table S4 and Figures S3–S5). The FeCO₂ significantly decreased with the increase in the preset F_DO₂ also during OM, HCM, and HOM patterns at constant PFR (p < 0.05 for all comparisons). Changes in the F_DO₂ significantly affected the mVR also during HCM pattern (p < 0.05). The VAS score decreased significantly with increasing F_DO₂ only at the PFR of 80 L/min during the CM (Table S4, Fig. S3C), OM (p < 0.001), and HOM (p = 0.004) patterns. All other variables during stable PFR were independent of F_DO₂ (p > 0.05 for all comparisons).

Discussion

Most of the studies on respiratory physiology during HFNC therapy in healthy volunteers and patients have focused on the EEP, breathing pattern changes, and work of breathing. The heterogeneous data for these parameters could be explained by heterogeneous sites of measurements: the placement of the tip of the probe in different parts of the volunteer’s upper airways (nasopharynx, nasal cavity, oropharynx, or hypopharynx) or inside the mannequin. We first focused on the gas composition in the hypopharynx and the influences of the combination of the PFR, breathing pattern, and F_DO₂ on FiO₂ (including oxygram shape), FeCO₂, EEP, and VTe.

FiO₂

Data on FiO₂ measurements inside airways close to the trachea or inside the trachea in healthy subjects and patients with ARF are limited. Similar FiO₂ values ​​were obtained in healthy volunteers, models, or patients, when the catheter was placed in the hypopharynx or inside the trachea (5–6), but the FiO₂ values measured in the nasopharynx was lower [7].

We conducted the first study in which the resulting FiO₂ in the hypopharynx at different respiratory patterns, PFR, and F_DO₂ values were evaluated. We found a very high range of FiO₂ within the combination of F_DO₂ at different PFR and breathing patterns. These results may call into question the accuracy of the measurement of the ROX index and SpO₂/FiO₂ during HFNC therapy in clinical studies, especially with respect to the OM and HOM patterns [16]. Sun YH et al. examined a lung model and observed a relationship between the PFR, F_DO2, and peak inspiratory flow, in which resulting FiO₂ was much lower than that in our data [17].

FiO₂ measurements in the hypopharynx during HFNC therapy with the OM, especially the HOM breathing pattern, are challenging due to “abnormal” oxygram waveforms during inspiration. The inverted oxygram may be a consequence of the Venturi effect with air entrainment through an open mouth and collision of the expiratory and inspiratory flows. Future studies are needed to assess the clinical relevance of these findings.

FeCO₂

Bench studies have shown an improvement in CO₂ clearance during HFNC therapy in the nasal cavity, pharynx, and trachea (8–10), which is more prominent with higher PFR and prolonged Te (decreased RR) [9]. Onodera Y et al. reported that CO₂ clearance during the OM pattern was higher than during the CM pattern, which is consistent with our data [10].

Although bench studies revealed a flow-dependent improvement in CO₂ clearance, the conclusions from volunteer studies were not as clear. Moller W et al. confirmed their previous bench data [8] on flow-dependent CO₂ clearance and clearance half-times using scintigraphy [18]. In addition, other studies did not identify changes in arterial, tissue, and capillary CO₂ levels (PaCO₂, PtCO₂, or PcCO₂, respectively), or FeCO₂ during HFNC [7, 19, 20]. If constant PaCO₂/PtCO₂/PcCO₂ levels could be explained by a combination of a reduced VD and a decrease in the RR during HFNC therapy and, as a consequence, stable VA, then a constant PaCO₂/PtCO₂/PcCO₂ during unchanged MV with increased VD washout and an increased VT could be explained only by an incorrect VT measurement or probable increased alveolar VD due to the CPAP-like effect with alveolar overdistension [20]. In our study, we measured FeCO₂, which is close to PaCO₂, and observed a decrease in FeCO₂ during constant CO₂ production, suggesting increased VA. The significant decrease in the RR with decreasing FeCO₂ in our study could be explained by a significant increase in the VT during HFNC therapy. We observed a pronounced effect on CO₂ washout, which was influenced by FiO₂ levels and the respiratory pattern of the subject (CM, OM, HCM, or HOM). To our knowledge, our study is the first to compare FeCO₂ during four different breathing patterns.

Numerous clinical studies of hypercapnic patients have confirmed data on improved CO₂ washout in patients, revealing a decrease in PaCO₂/PtCO₂/PcCO₂ [21–30]. However, some studies on AHRF did not find a decrease in PaCO₂/PtCO₂/PcCO₂, which could be explained by constant VA in these patients or increased CO₂ production [31, 32].

The following question remains unanswered: does end-tidal CO₂ during HFNC therapy reflect only alveolar gas or mixed gas from alveoli and entrained gas from high-flow cannulas?

VT

High tidal volumes during HFNC may injure the lungs through lung overinflation (strain) or patient self-inflicted lung injury (P-SILI) [33, 34]. VT above 9.5 mL/kg were found to be a predictor of NIV failure in a single multicentre observational study of patients with moderate-to-severe hypoxemic ARF treated with NIV [35].

The VT measurement methods used during HFNC therapy in published studies are not ideal. These factors may lead to conflicting results showing the opposite direction of VT changes with an increase in the inspiratory flow rate. We subdivided the discussion of VT changes during HFNC therapy by the VT measurement method.

Data on VT measurements during HFNC therapy obtained using electrical impedance tomography [31, 32, 36–39] and inductive plethysmography [19, 21, 40] have shown various results in healthy volunteers and patients with ARF. Studies using thoracic bioimpedance (ExSpiron 2Xi, Senzime, Sweden) in healthy volunteers [20] and volumetric measurements during HFNC interruption in patients with ARF [41] showed an increase in the VT of up to 11 ml/kg IBW, which is consistent with our data.

To the best of our knowledge, a time-of-flight camera is the most accurate noninvasive method for VT measurements in nonintubated patients. A single study used this method to assess VT during noninvasive respiratory support [42]. This study revealed that an increase in the PFR and/or an increase in the subject’s respiratory drive was associated with an increase in VT to 913 (680–1166) mL. Moreover, with a PFR of 30 L/min, 50% of the subjects had a VT greater than 10 mL/kg, and 15% had a VT > 15 mL/kg; however, with a PFR of 60 L/min, 65% of the subjects had a VT greater than 10 mL/kg, 35% had a VT greater than 15 mL/kg, and 10% had a VT greater than 20 mL/kg. These results are consistent with our data.

End-expiratory pressure and end-expiratory volume

Bench studies have shown highly heterogeneous values for EEP during the CM phase, ranging from 0–3 cm H₂O [38] when measured inside a mannequin “nasopharynx” to 16 cm H₂O at a PFR of 60 L/min when measured inside a lung model [10, 17]. Substituting appropriate numbers from our study during the CM pattern to their equation, we obtain an EEP level equal to 14.5 cmH₂O inside the lung model, which corresponds to our data. In our study, we observed a long phase of positive EEP, which contradicts the findings of bench studies that documented only a short (“peak”) EEP [17].

Volunteer studies also revealed heterogeneous data for EEP values. Most studies have measured the EEP in the nasopharynx or distal part of the nasal cavity. These studies reported concordant results: EEP levels were approximately the PFR divided by ten [38, 43–45]. In contrast, two studies that measured EEP in the oropharynx [46] or the hypopharynx [7] reported higher EEP values during the CM pattern, which were close to our data but presented high variability [7]. PEEP levels also depend on the cannula size: the EEP level was higher with a 5 mm cannula than with a 4 mm or 3 mm cannula [47]. Moreover, the volunteers in the Chinese study [48] had higher PEEP levels than those in the European study did, possibly due to the anatomical features of the upper airways.

Clinical studies using EIT revealed an increase in end-expiratory lung impedance (EELI) in all studies [22, 32, 33, 37, 39, 48], with greater EELI at higher PFR. Surprisingly, in two clinical studies, the EEP was lower than that in volunteer studies (including our study). This difference can be explained by partial mouth opening during HFNC therapy in these patients [6, 37]. Villalba DS have found no differences between tracheal and pharyngeal pressures during HFNC with different PFR during the CM and OM patterns in patients with a tracheostomy during the process of decannulation [12]. The clinical data on EEP are discussed in detail in a dedicated review [49].

Subject’s comfort and work of breathing (WOB)

In the previous studies, the WOB in healthy subjects was not significantly different at the range of PFR of 10–40 L/min but increased significantly at higher PFR or when larger cannulas were used [19, 38, 47]. In healthy volunteers, the WOB was higher during HFNC therapy at 40 and 60 L/min during the CM pattern than during CPAP therapy at 4 cmH₂O for a single breath, also the WOB during the CM pattern was greater than that during the OM pattern [38]. In contrast, we found that about half of the volunteers preferred the CM pattern as more comfortable. In AHRF patients, dyspnea (Borg score) was less pronounced during HFNC therapy than during low-flow oxygen therapy [37] but was more pronounced if excessive PFR were used, and it was significantly higher if the PFR was 1.5 L/min per ideal body weight [48], which is comparable to our VAS data obtained at a PFR of 80 L/min in healthy volunteers.

Our study has several strengths: 1. simultaneous monitoring of capnogram, oxygram, pressure, and volume measurements and VAS comfort during HFNC therapy; 2. comparisons of three different PFR, including an unusually high flow rate of 80 L/min; 3. The use of combinations of different F_DO₂ values and PFR to reveal the influences of these combinations on the gas composition in the hypopharynx; 4. measurements were made in the hypopharynx, close to vocal cords, which made it more reliable for clinical practice; 5. The use of different breathing patterns, including not only CM and OM but also hyperpnea during CM and OM; and 6. The calculation of the mVR at all preset combinations of flow rates, F_DO₂ values, and breathing patterns.

The limitations of the study include the following: 1. The VT measurement technique may introduce some bias due to the Venturi effect with gas entrainment and changes in the breathing pattern during the VT measurement; we mitigated this bias by training volunteers in the optimal measuring technique and excluded subjects with abnormal oxygrams in whom the Venturi effect was evident from the analysis. 2. The results may not be fully translated into clinical practice for patients with ARF due to pathophysiological changes.

Conclusions

HCM, OM, and HOM breathing patterns decreased FiO₂ and F_ECO₂ in the hypopharynx of healthy subjects (which was more pronounced during the OM and HOM patterns). The FiO₂ level in the hypopharynx during the OM and HOM patterns is unpredictable at all PFR. A PFR of 60 L/min was sufficient to maintain FiO₂ close to F_DO₂ during the CM and HCM patterns. HFNC within the CM and HCM provided flow-dependent CPAP-effects over a wide range and could be associated with lung hyperinflation. An excessive PFR led to discomfort.

Abbreviations

AHRF

Acute hypoxemic respiratory failure

CM

Closed mouth,

COPD

Chronic obstructive pulmonary disease

CO₂

Carbon dioxide

CPAP

Constant airway pressure

DeltaP

Delta of pressure

EELI

End-expiratory lung impedance

EEP

Measured end-expiratory pressure in the hypopharynx

Efmean

Mean expiratory flow

F_DO₂

Preset oxygen fraction on HFNC device

F_ECO₂

Measured expiratory carbon dioxide fraction in the hypopharynx

FiO₂

Measured inspiratory oxygen fraction in the hypopharynx

HCM

Combination of CM with hyperpnea

HFNC

High flow through nasal cannula

HOM

Combination of OM with hyperpnea

IBW

Ideal body weight

IQR

Interquartile range

MV

Minute ventilation

OM

Open mouth

O₂

Oxygen

PaCO₂

Arterial carbon dioxide partial pressure

PcCO₂

Capillary carbon dioxide partial pressure

PEEP

Positive end-expiratory pressure

PetCO₂

End-expiratory carbon dioxide partial pressure

P-SILI

Patient self-inflicted lung injury

PtCO₂

Tissue carbon dioxide partial pressure

ROX index

Calculated as SpO₂/FiO₂/RR

RR

Respiratory rate

SpO₂

Peripheral oxygen saturation

Te

Expiratory time

Ti

Inspiratory time

VAS

Visual-analog scale

V’CO₂

CO₂ production,

VD

Dead space volume

VR

Ventilatory ratio

VT

Tidal volume

Vte

Exhaled measured tidal volume

WOB

Work of breathing

Authors’ contributions

A.I.Y.: study design, data collection, analysis, manuscript writing, and revision; А.P.K.: data collection and analysis; F.D.T.: data collection; A.V.R.: data collection; D.D.Z.: data collection; V.D.R.: data analysis and interpretation; M.I.S: data collection; M.M.N.: data collection; Z.M.M.: data collection; S.N.A.: manuscript editing and revision. All authors revised the drafted manuscript, and all read and approved its final version.

Funding

No funding source.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Institutional Ethics Committee of Sechenov University (reference number: 25–23, date of approval 14/12/2023) and registered on ClinicalTrials.gov (NCT06189716). All methods were performed in accordance with the Declaration of Helsinki and the international ethical guidelines for human biomedical research. Written informed consent was obtained from all healthy volunteers enrolled in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.