Effectiveness of High-Flow Nasal Cannula on Pulmonary Rehabilitation in Subjects with Chronic Respiratory Failure

Backgrounds: There are no clinical data comparing the effect of exercise training using both high fraction of inspired oxygen (F I O 2 ) and high ow through a high-ow nasal cannula (HFNC) with that of ordinary supplemental oxygen on exercise capacity in subjects with chronic respiratory failure (CRF) receiving long-term oxygen therapy (LTOT). Methods: In this randomized study, 32 patients with CRF receiving LTOT were assigned to undergo 4 weeks of exercise training on a cycle ergometer using an HFNC (ow: 50 L/min) with a F I O 2 of 1.0 (HFNC group; n=16) or ordinary supplemental oxygen via a nasal cannula (ow: 6 L/min) (oxygen group; n=16). Before and after 4 weeks of exercise training, a 6-min walking test, constant-load test, and blood sampling were performed. Results: Prior to exercise training, the endurance time of the constant-load exercise test using an HFNC was signicantly longer than that reported with prescribed oxygen in daily LTOT (p=0.004) or a 6 L/min nasal cannula (p=0.0003). Following 4 weeks of exercise training, change in the 6-min walking distance was signicantly greater in the HFNC versus the oxygen group (55.2±69.6 m vs. —0.5±87.3 m, respectively; p=0.04). The plasma levels of adrenaline, noradrenaline, and serum C-reactive protein were signicantly decreased only in the HFNC group, after 4 weeks of exercise training. Conclusions: Despite heterogeneity in the effect among patients, exercise training using both high F I O 2 and high ow through an HFNC is a potentially superior exercise training modality for CRF patients receiving LTOT.

It has been observed that patients with advanced chronic respiratory failure (CRF) receiving long-term oxygen therapy (LTOT) have severe hypoxia during exercise training even if they receive relatively high ow rates of oxygen via a nasal cannula. In addition, the maximal ow rate of oxygen via a nasal cannula during exercise training is approximately 6 L/min. This is due to severe nasal dryness and pain induced by higher ow rates. Low-intensity exercise training has been applied to individuals with CRF who have di culty in achieving their target intensity because of dyspnea and/or hypoxia. However, it led to insu cient exercise training effects compared with high-intensity exercise training.
Several new options to increase the effects of exercise training have been reported. It has been shown that hyperoxia during exercise training increased the peak work rate [8] or endurance time in the constantload exercise test [9] in COPD patients. In other studies, exercise training using hyperoxia did not demonstrate a bene cial effect [10][11][12][13]. The administration of a high ow heated and humidi ed gas mixture through high-ow nasal cannula (HFNC) promotes higher and more stable inspiratory oxygen fraction values, decreases anatomical dead space, and generates a positive airway pressure that can improve ventilatory e ciency [14][15][16]. In addition, it reduces the effort of breathing and enhances patients' comfort and tolerance [16][17][18][19]. In a previous study, the duration of the constant-load exercise test in COPD patients receiving LTOT was signi cantly improved after exercise training consisting of 20 supervised sessions in both the HFNC group and control group, whereas no signi cant difference between groups was observed in improvement in the duration of the constant-load exercise test [20]. In this study, both groups undergone exercise training at the same fraction of inspired oxygen (F I O 2 ). Based on the bene cial effects of an HFNC, we postulated that high intensity exercise training using not only high ow but also high F I O 2 through an HFNC may lead to an enhancement of exercise capacity in patients with advanced CRF receiving LTOT.
The aim of this randomized, prospective study was to compare the effect of 4 weeks of exercise training using both high F I O 2 and high ow through an HFNC or supplemental oxygen via a nasal cannula on the exercise capacity (e.g., 6-min walking distance [6MWD]) of CRF patients receiving LTOT.

Patients
From June 2016 to May 2018, patients with CRF (aged 40-90 years) undergoing LTOT for ≥ 3 months and attending an in-patient rehabilitation program were eligible to participate in the study. Patients with a stable clinical condition (i.e., no exacerbation reported for 3 months prior to the study) and absence of the following exclusion criteria: history of cardiovascular disease (e.g., myocardial infarction or ischemic heart disease, diabetes mellitus under treatment with hypoglycemic agents or insulin, neurological disorders, renal dysfunction (serum creatinine ≥ 1.2 mg/dL), and unable to undergo rehabilitation were enrolled.
In this prospective study, we assessed 42 consecutive patients with CRF receiving LTOT. Prior to enrollment, all patients underwent a 2-week screening period. Of those, seven patients were excluded (two with acute exacerbation, three with history of cardiovascular disease, and two who were unable to undergo rehabilitation), and three patients declined to participate. Thus, the remaining 32 clinically stable CRF patients receiving LTOT were enrolled (Clinical Trial Registration-URL: http://www.clinicaltrials.gov. Unique identi er: NCT02804243). This study was approved by the Ethics Committee of the National Hospital Organization Minami-Kyoto Hospital. All patients provided written informed consent prior to their participation.

Study design
On day 1, arterial blood gases (ABGs) were obtained in the supine position under a prescribed ow of oxygen via a nasal cannula and patients underwent a 6-min walking test (6MWT). On day 2, venous blood samples were obtained in the fasting state in the morning and examined for C-reactive protein (CRP), and plasma catecholamine levels. Pulmonary function tests were performed using the CHESTAC system (Chest M.I. Inc., Tokyo, Japan). In addition, patients underwent an incremental-load exercise test under prescribed oxygen on a cycle ergometer (Aerobike75XL, COMBI, Tokyo, Japan) to assess their maximal exercise capacity. In the following 3 days, the patients underwent (in random order) three constant-load exercise tests with a prescribed ow of oxygen via a nasal cannula, a ow of 6 L/min of oxygen via a nasal cannula, or an HFNC at a ow of 50 L/min with F I O 2 of 1.0. The HFNC delivered a heated and humidi ed gas through an Opti ow system (Fisher and Paykel Healthcare, Auckland, New Zealand) using large-bore bi-nasal prongs.
After the constant-load exercise tests, patients completed a 4-week period of exercise training ( ve sessions per week) using the same workload achieved during the constant-load exercise tests. In the HFNC group, the HFNC was set at maximum F I O 2 (F I O 2 of 1.0) and near-maximum inspiratory ow rate (50 L/min) to assess its maximum effects. In the oxygen group, the ow rate of oxygen via a nasal cannula was set at the maximal ow rate (6 L/min).
Venous blood sampling, ABGs, pulmonary function tests and 6MWT on prescribed oxygen were conducted after 4 weeks of exercise training. In the following 3 days, subjects also underwent (in random order) three constant-load exercise tests using the same workload with that of the baseline assessment with prescribed oxygen in daily LTOT, a 6 L/min nasal cannula, or an HFNC at F I O 2 of 1.0 and 50 L/min.
The primary outcome was change in the 6MWD prior to and following 4 weeks of exercise training. The secondary outcome was change in the duration of constant-load exercise test prior to and after 4 weeks of exercise training.

6MWT
The 6MWT was performed under prescribed oxygen in daily LTOT. According to the guidelines of the American Thoracic Society [21], patients were encouraged using standard methodology every minute of the 6MWT. A pulse oximeter (Anypal Walk, Fukuda Denshi, Tokyo, Japan) with a nger probe was used to measure oxygen saturation by pulse oximetry (SpO 2 ) and the pulse rate during the 6MWT. In addition, a modi ed Borg scale [22] was used to quantify the levels of dyspnea perceived by patients at each minute during the 6MWT.

Exercise testing
Both incremental-and constant-load exercise tests were performed with continuous monitoring of SpO 2 and the pulse rate on a cycle ergometer. In addition, continuous transcutaneous carbon dioxide partial pressure (PtcCO 2 ) monitoring was performed using a TOSCA measurement system and TOSCA 500 monitor (Radiometer, Copenhagen, Denmark) [23,24]. After automated calibration, the TOSCA monitor was attached to the right or left ear lobe. PtcCO 2 monitoring was initiated after at least 10 min of equilibration. The incremental-load exercise test was performed according to the appropriate Japanese guidelines [25]. The incremental-load exercise test under prescribed oxygen in daily LTOT was initiated at 5 watts and increased by 5 watts every 2 minutes until the patient's SpO 2 was ≤ 85%, the pulse rate was ≥ 135 beats/min, or the modi ed Borg scale for dyspnea was ≥ 5. Prior to and following 4 weeks of exercise training, the patients performed three constant-load exercise tests at 80% of the maximum workload achieved with the incremental-load exercise test, with prescribed oxygen in daily LTOT, 6 L/min via a nasal cannula, or an HFNC at F I O 2 of 1.0 and 50 L/min. The criteria for the termination of the constant-load exercise test were the same as those for the incremental-load exercise test. The SpO 2 , pulse rate, and PtcCO 2 were measured at 6 min of the constant-load exercise test. All exercise testing was performed in the afternoon.

Exercise training
The patients performed supervised pulse oximeter-monitored continuous exercise training (5 sessions per week) receiving 6 L/min using a nasal cannula or an HFNC at F I O 2 of 1.0 and 50 L/min. All training sessions were conducted on the same cycle ergometer and supervised by medical staff. Initially, the workload of the exercise training was the same as that of the constant-load exercise test. The criteria for the termination of exercise training sessions were SpO 2 ≤85%, pulse rate ≥135 beats/min, or modi ed Borg scale for dyspnea ≥5. Whenever the duration of an exercise training session was ≥30 minutes, the workload of the subsequent session was increased by 5 watts.

Randomization
This study was randomized via blinded envelope prior to the initiation of the study.

Power analysis
Based on a previous study [26], differences of 30 meters of change in the 6MWD prior to and following 4 weeks of exercise training between the HFNC group and the oxygen group were decided to be clinically signi cant, and a standard deviation of 30 meters was expected. The sample size was set to achieve 80% power at a 5% signi cance level. The calculated sample size in each group was 16 subjects.

Statistical analysis
Data were analyzed using the JMP 9.0 software (SAS Institute, Inc. Cary, NC, USA), and values are expressed as mean ± standard deviation or absolute numbers and percentages. We compared subject characteristics, the results of the 6MWT and constant-load exercise test, and other parameters between the HFNC and oxygen groups. Continuous variables were tested using the unpaired t test or Mann-Whitney U test. Categorical variables were compared using the χ 2 test or Fisher's exact test. To investigate changes in the 6-min walk distance, constant load exercise test, and other parameters prior to and after 4 weeks of exercise training in each group, comparisons of data between those two time points were tested using a paired t test. A p < 0.05 denoted statistical signi cance.

Results
Patient characteristics Figure 1 shows the study ow chart of this randomized study. In this study, 32 CRF patients receiving LTOT were randomly assigned to undergo 4 weeks of exercise training using an HFNC (n = 16) or a 6 L/min nasal cannula (n = 16). The characteristics of the patients are summarized in Table 1. The cause of CRF was COPD (n = 13), interstitial pulmonary brosis (IPF) (n = 15), or bronchiectasis (n = 4). There were no signi cant baseline differences between the two groups (Table 1). Similarly, 6MWD, the maximum workload or time of the incremental-load exercise test, and time of the constant-load exercise test under three conditions did not differ signi cantly between the two groups ( Table 2).  Figure 2 shows the results of the constant-load exercise test prior to exercise training in all patients. The duration of the constant-load exercise test using an HFNC was signi cantly longer than that reported with prescribed oxygen in daily LTOT (p = 0.004) or a 6 L/min nasal cannula (p = 0.0003) (Fig. 2-A). Moreover, there was no signi cant difference between the two conditions (prescribed oxygen in daily LTOT or a 6 L/min nasal cannula) in the duration of the constant-load exercise test (p = 0.15). The modi ed Borg scale for dyspnea at the termination of the constant-load exercise test under the HFNC tended to be lower compared with that observed under the 6 L/min nasal cannula (p = 0.07) (Fig. 2-B).
Six minutes after initiation of the constant-load exercise test, the SpO 2 was signi cantly higher and the pulse rate was signi cantly lower with HFNC versus those observed with prescribed oxygen in daily LTOT or 6 L/min nasal cannula ( Fig. 3-A, 3-B). There were no signi cant differences between the three conditions in PtcCO 2 six minutes after initiation of the constant-load exercise test ( Fig. 3 Table 3). Compared with the baseline values, the 6MWD was signi cantly increased in the HFNC group, but not in the oxygen group (Table 4, Fig. 4). Notably, change in the modi ed Borg scale for dyspnea at the termination of the 6MWT was not signi cantly different between the two groups (Table 3).   (Table 4). In contrast, the duration of the constant-load exercise test under prescribed oxygen in daily LTOT in the HFNC group increased signi cantly compared with that observed at baseline (p = 0.01). However, this increase was not observed in the oxygen group (p = 0.10) ( Table 4).

Effects of 4 weeks of exercise training on variables
Regarding changes in the pulmonary function tests, ABGs, the plasma levels of adrenaline, noradrenaline, and serum CRP prior to and following 4 weeks of exercise training, the plasma levels of adrenaline, noradrenaline, and serum CRP were signi cantly decreased in the HFNC group compared with the respective baseline values. However, these effects were not observed in the oxygen group (Table 4). ABGs and pulmonary function did not worsen after 4 weeks of exercise training using not only 6L/min oxygen but also HFNC (both extremely high F I O 2 and extremely high ow).

Discussion
The present study was the rst randomized, prospective study demonstrating that the use of both high F I O 2 and high ow through an HFNC signi cantly prolonged the duration of the constant-load exercise in CRF patients receiving LTOT versus prescribed oxygen in daily LTOT or a 6 L/min nasal cannula.
Moreover, this study found that 4 weeks of training using both high F I O 2 and high ow through an HFNC signi cantly improved the 6MWD compared with using a 6 L/min nasal cannula. Moreover, only patients using an HFNC exhibited a signi cant decrease in the plasma levels of adrenaline, noradrenaline, and serum CRP. These results suggest that exercise training using both high F I O 2 and high ow through an HFNC may be a preferable modality versus a nasal cannula in CRF patients receiving LTOT.
The American Thoracic Society/European Respiratory Society statement on eld walking tests determined an increase of ≥ 30 meters in the 6MWD (with a variability of 25 to 33 meters) as clinically relevant [27]. In the present study, 4 weeks of exercise training using an HFNC improved the 6MWD by 55 meters. Although the sample size in this study was small, the results can be considered signi cant and de nitive.
Several factors for improving the duration of exercise using an HFNC have been considered. Firstly, the F I O 2 values induced via an HFNC are more stable and much higher than those of standard oxygen delivery systems [28]. Use of an HFNC can achieve a F I O 2 of 1.0, whereas the F I O 2 associated with a 6 L/min nasal cannula was estimated to be approximately 0.4. Additionally, an HFNC generates a high ow rate that can exceed the subject's peak inspiratory ow rate, thus reducing entrainment of room air and dilution of the administered oxygen [29,30]. Secondly, continuous ushing of the upper airway via an HFNC reduces dead space [31]. This effect may enhance alveolar ventilation if tidal volume is the same, and should increase the oxygen concentration of upper airway at the end of expiration. Thirdly, the high ow rates of an HFNC generate a positive nasopharyngeal pressure which linearly correlates with the administered ow rate in healthy volunteers and patients with stable COPD, IPF, and postcardiac surgery [29,[32][33][34][35][36]. When the mouth of the patients is open, the HFNC produces low positive nasopharyngeal pressure [35] and this condition is frequent during exercise. This low-level positive nasopharyngeal pressure generated using an HFNC may reduce the effort of breathing through enhancing oxygenation by positive end-expiratory pressure (PEEP), providing a low-level inspiratory assistance, and reducing a preload as a counter-PEEP effect. We assume that the use of an HFNC enabled patients to perform prolonged exercise training through the aforementioned mechanisms, resulting in a signi cant improvement in the 6MWD.
In the present study, our patients had advanced CRF. Thus, their exercise capacity was presumed to be low. To exert the maximal effect of exercise training on exercise capacity, we assessed the maximal effects of an HFNC on long-term exercise training using both maximum F I O 2 and near-maximum inspiratory ow rate. The present study showed that 4 weeks of training using an HFNC signi cantly increased the 6MWD. However, it is unclear whether high ow rate or 100% FiO 2 are responsible for the bene cial effects on exercise capacity. Future studies comparing the effect of long-term exercise training using supplemental oxygen or HFNC with the equivalent F I O 2 on the exercise capacity are warranted.
In this study, the HFNC group was used a F I O 2 of 1.0 during exercise training. Prolonged hyperoxia has been implicated in organ toxicity processes, such as acute lung injury [37][38][39]. Systemically, hyperoxia induces peripheral vasoconstriction [40] and, increases production of reactive oxygen species [41].
Recently, arterial hyperoxia is associated with poor hospital outcome in various subsets of critically ill patients [42][43][44]. On the other hand, there are few studies on the effects of high concentration oxygen administration during exercise training. It has been demonstrated that 100% oxygen administration during 8 weeks of aerobic high-intensity interval training increased peak oxygen uptake and peak workload to a considerable extent in severe COPD patients [45]. In our study, the HFNC group showed no deterioration in PaO 2 values, pulmonary function tests including diffusion capacity of the lung for carbon monoxide, and the modi ed Borg Scale for dyspnea on 6MWT (Table 4). These results indicate that 4 weeks of exercise training using an HFNC at F I O 2 of 1.0 might not induce hyperoxic lung injury. Indeed, the possibility of a much longer period of training using an HFNC with high F I O 2 inducing hyperoxic lung injury is undeniable.
In addition, the use of a high F I O 2 during exercise for CRF patients may cause CO 2 retention. In the present study, the partial pressure of arterial carbon dioxide values in the HFNC group were not signi cantly changed prior to and after 4 weeks of exercise training (Table 3). In addition, there were no signi cant differences between the three conditions (i.e., prescribed oxygen in daily LTOT vs. 6 L/min via a nasal cannula vs. HFNC at F I O 2 of 1.0 and 50 L/min) in PtcCO 2 6 min after initiation of the constantload exercise test ( Fig. 3-C). Based on these results, we assume that the risk of CO 2 retention caused by exercise training using an HFNC at FIO 2 of 1.0 is low.
In this study, the plasma levels of adrenaline, noradrenaline, and serum CRP were examined to investigate the effects of exercise training on systemic in ammation and sympathetic activity. The results showed that the HFNC group had a signi cant reduction in the plasma levels of adrenaline, noradrenaline, and serum CRP after 4 weeks of exercise training compared with the baseline values. It is likely that exercise training reduces the level of CRP by decreasing the production of cytokines in adipose tissue, skeletal muscles, endothelial and blood mononuclear cells, improving endothelial function and insulin sensitivity, and inducing an antioxidant effect [46]. Previous studies reported that the observed enhancement of antioxidant pathways and suppression of pro-oxidant mechanisms in the rostral ventrolateral medulla of rabbits with chronic heart failure contribute to the normalization of sympathetic nerve activity after exercise training [47]. It has been demonstrated that exercise training decreased the level of CRP in the serum [48,49] and sympathetic activity [50] in patients with COPD. Prolonged exercise training using an HFNC may lead to a decrease in the plasma levels of adrenaline, noradrenaline, and serum CRP.
This study had several limitations. Firstly, the underlying disease in the patients of this study was heterogeneous, so both high F I O 2 and high ow through an HFNC may have different effects on their underlying disease. In addition, their respiratory failure was severe (thus receiving LTOT). Therefore, it may be di cult to generalize the results of the present study to other patient populations such as those with COPD alone, mild respiratory failure, etc. Secondly, the sample size of this study was small. However, the differences observed in the improvement of the 6MWD after exercise training between the HFNC and oxygen groups were su ciently large to be considered signi cant. Thirdly, the oxygen group did not demonstrate a signi cant improvement in the 6MWD after 4 weeks of exercise training. Respiratory failure in our patients was severe; thus, we assumed that they may be at a high risk of adverse events, such as severe hypoxia and arrythmia during exercise training. Therefore, we set strict criteria for the termination of the exercise training (ie. SpO 2 ≤ 85%, pulse rate ≥ 135 beats/min) to prevent the occurrence of adverse events during the test. Although there were no adverse events observed during exercise training in the present study, it is possible that the oxygen group had a shorter exercise training duration owing to more frequent stoppage of the training in response to desaturation or tachycardia. Therefore, signi cant improvement in the 6MWD after 4 weeks of exercise training may not be observed in the oxygen group.

Conclusion
We demonstrated that 4 weeks of exercise training using both high F I O 2 and high ow through an HFNC signi cantly improved the 6MWD, and decreased systemic in ammation and sympathetic activity. Future research involving large sample sizes or homogenous populations (e.g., patients with COPD alone) is warranted. However, based on the present ndings, exercise training using both high F I O 2 and high ow through an HFNC is safe and may become an effective modality in CRF patients receiving LTOT.

Declarations
Ethics approval and consent to participate This study was approved by the Ethics Committee of the National Hospital Organization Minami-Kyoto Hospital. All patients provided written informed consent prior to their participation.

Consent to publish
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Competing interests
Drs. Yuichi Chihara, Tomomasa Tsuboi, Kensuke Sumi, and Atsuo Sato have no con icts of interest to disclose.

Funding
This study was supported by Fukuda Foundation for Medical Technology. This nancial support was used to purchase SpO 2 monitors and TOSCA sensors. Abbreviations: LTOT, long-term oxygen therapy; HFNC, high-ow nasal cannula; SpO2, percutaneous oxygen saturation; PtcCO2, transcutaneous carbon dioxide partial