Nasal High Flow Reduces Dead Space

: 39 Recent studies show that nasal high flow (NHF) therapy can support ventilation in patients 40 with acute or chronic respiratory disorders. Clearance of dead-space has been suggested as 41 being the key mechanisms of respiratory support with NHF therapy. 42 The hypothesis of this study was that NHF in a dose-dependent manner can clear dead space 43 of the upper airways from expired air and decrease re-breathing. 44 The randomized cross-over study involved 10 volunteers using scintigraphy with 81m Krypton- 45 gas ( 81m Kr-gas) during a breath-holding maneuver with closed mouth and in three nasally 46 breathing tracheotomized patients by volumetric capnography and oximetry through 47 sampling CO 2 and O 2 in the trachea and measuring the inspired volume with inductance 48 plethysmography following NHF rates of 15, 30 and 45 L/min. 49 The scintigraphy revealed a decrease in 81m Kr-gas clearance half-time with an increase of 50 NHF in the nasal cavities (cc = -0.55, p < 0.01), pharynx (cc = -0.41, p < 0.01) and the trachea 51 (cc = -0.51, p < 0.01). Clearance rates in nasal cavities derived from time constants and MRI- 52 measured volumes were 40.6 (SD 12.3), 52.5 (SD 17.7) and 72.9 (SD 21.3) mL/s during NHF 53 (15-30-45L/min). Measurement of inspired gases in the trachea showed an NHF-dependent 54 decrease of inspired CO 2 that correlated with an increase of inspired O 2 (cc = -0.77, p < 0.05). 55 NHF clears the upper airways from expired air, which reduces dead space by a decrease of 56 re-breathing making ventilation more efficient. The dead-space clearance is flow and time- dependent and it may extend below the soft palate. Part of the study has been registered at www.clinicaltrials.gov (NCT01509703).

although how NHF produces these effects is not yet understood. A mechanistic study on 90 healthy volunteers suggested two different ventilatory responses to NHF, one when awake 91 and another during sleep (19). In this study it was speculated that the reduction of dead-92 space ventilation due to clearance of anatomical dead-space in the upper airways could be 93 the principal driver for the reduction of minute ventilation during sleep, which may 94 potentially lead to a reduction in the work of breathing. In a previous study using upper 95 airway models the authors demonstrated the fast-occurring flow dependent clearance of 96 nasal cavities by NHF (18). The dead-space clearance is difficult to study in vivo due to the 97 complexity in quantifying the respiratory gases in the airways. However, many have 98 proposed it to be the major physiological mechanism, which improves respiratory support 99 (20, 22, 26) and reduces arterial and tissue CO 2 (1, 7, 14). 100

101
The aim of this study was to measure upper airway dead-space reduction during NHF 102 therapy to test a hypothesis that NHF in a dose-dependent manner can clear dead space in 103 Clearance of 81m Kr tracer gas from the upper airways by NHF was assessed in healthy 106 volunteers using dynamic gamma camera imaging. Reduction of re-breathing was 107 investigated in tracheotomized patients using volumetric capnography and oximetry by 108 For these experiments the 81m Kr-gas was generated and a planar gamma camera was used 136 for imaging, as described in detail earlier (18). The volunteers filled their upper airways with 81m Kr tracer gas through the nasal pillow, and the NHF cannula with the preset flow was 138 inserted into the nose while the volunteer was holding their breath. 81m Kr-gas activity-time 139 profiles were assessed in five regions of interest (ROI): anterior nasal (Nasal1), posterior 140 nasal (Nasal2), pharynx (space from the soft palate to the larynx), trachea and the upper 141 lung ( Figure 1A). 81m Kr-gas clearance time constants and half-times were evaluated after 142 correction with the natural 81m Kr-gas decay (T 1/2 = 13 s). Nasal clearance rates were 143 evaluated as the ratio of nasal volume (V N ) and clearance time constant. Nasal volume, 144 comprising the nasal cavity and the nasopharynx (excluding sinuses) was assessed using 145 individual MRI imaging. 146 147

Clearance of anatomical dead space in tracheotomized patients 148
Tracheotomized patients were included in order to assess re-breathing of expired gas from 149 the upper airways. When the weaning from invasive mechanical ventilation was completed 150 the tracheostomy tube was replaced with a tracheostomy retainer (2). A custom-made 151 probe was placed through the retainer to measure O 2 , CO 2 and pressure profiles for 152 synchronization with breathing (ADInstruments, New Zealand). Inspiratory volume was 153 assessed with calibrated respiratory inductance plethysmography (RIP; Viasys Services, USA), 154 as described in detail previously (12,19). 155

156
The effect of NHF on the volume of inspired O 2 and CO 2 was analyzed for every breath. 157 Inspired O 2 was calculated in the first 100 mL of inspired volume. Inspired CO 2 was 158 calculated in the total inspired volume and in the first 100 mL. Arterial blood oxygen 159 saturation (SpO 2 ) and transcutaneous CO 2 (Tosca, Radiometer, Denmark) were monitored 160 throughout the study. 161 162

Data analysis 163
All data is presented as mean +/-standard deviation (SD). Differences between groups or 164 application modes were assessed by a two-sided t-test using a significance level of p < 0.05. 165 Pearson's coefficient correlation (cc) analysis was then applied, to assess the correlation 166 among the study variables. After filling the upper airways with 81m Kr-gas the volunteer was holding his or her breath and 172 the NHF cannula was attached to their nose; this caused immediate purging of the 81m Kr-gas 173 from the upper airways ( Figure 1B and supplemental video). NHF caused rapid activity decay 174 in the nasal cavity and, as shown in Figure 1B, the nasal cavity was cleared at 0.5 s after 175 applying NHF at a rate of 45 L/min. 176

177
The half-times of 81m Kr-gas clearance in nasal regions are shown in Table 2 and Figure 2A. respectively. This demonstrates that there is a significant correlation between clearance rate 186 and NHF (cc = 0.61, p < 0.01). 187

188
In the lower compartments beyond the soft palate, 81m Kr-gas clearance was also NHF 189 dependent but slower (pharynx: cc = -0.41, p < 0.01; trachea: cc = -0.51, p < 0.01; Table 2 and 190 Figure 2B) and in some experiments only natural 81m Kr-gas decay was recorded. Pharyngeal 191 and tracheal clearance half-times correlated with the nasal half times (cc = 0.4, p < 0.05). 192 There was no detected 81m Kr-gas clearance in the lung ROI. 193 194

Re-breathing of expired air during NHF therapy in tracheotomized patients 195
An example of a single-breath analysis of inspired CO 2 and O 2 at baseline and during an NHF 196 rate of 45 L/min is presented in Figures 3A and 3B. A summary of the effects of NHF on 197 inspired CO 2 and O 2 in the first 100 mL is shown in Figure 4. In all three patients studied, NHF 198 led to a decrease of inspired CO 2 and to an increase of inspired O 2 in a flow-dependent 199 manner ( Figure 4A and 4B). Linear regression analyses between a change (Δ) of total inspired 200 O 2 versus CO 2 in the first 100 mL per breath are presented in Figure 4C. An NHF-induced 201 decrease of inspired CO 2 correlates with an increase of inspired O 2 (cc = -0.767; r 2 = 0.59, p = 202 0.016). A ratio between inspired CO 2 in the first 100 mL of inspired volume to the total 203 inspired CO 2 grouped by all baselines and NHF treatments is presented in Figure 4D. NHF 204 resulted in a significantly higher ratio during NHF treatment relative to baseline ventilation 205 (0.84 (SD 0.10) vs. 0.75 (SD 0.12); p < 0.01, paired t-test). Change of tidal volume, respiratory 206 rate, minute ventilation as well as SpO 2 and tissue CO 2 throughout the study are presented 207 in Table 3. 208 209 210

212
In the first part of the study, dead-space clearance by NHF therapy was analyzed in 10 213 healthy volunteers by the use of 81m Kr-gas, a radioactive tracer gas and a gamma camera. 214 The major findings in this investigation are the NHF-dependent reduction of radioactive 215 tracer-gas clearance half-times in the upper airways with very fast removal of the tracer gas 216 from the nasal cavities (half-times < 0.5 s at an NHF rate of 45 L/min) that confirmed the 217 authors' model study (18). Further in various volunteers significant 81m Kr-gas clearance was 218 detected in deeper compartments below the soft palate, which could be investigated only in 219 vivo. Rates of NHF in the range of 15 to 45 L/min were used, which were also used previously 220 (18) and which is common in clinical settings for adults. NHF rates up to 60 L/min were used 221 in patients with acute respiratory failure (28), but cannot be well tolerated by some naïve 222 healthy participants that were found during the preparation of the experiments. In the 223 second part of the study, tracheal O 2 and CO 2 breathing profiles in three tracheotomized 224 patients revealed an NHF-dependent increase of inspired O 2 and a decrease of inspired CO 2 , 225 which confirmed a reduction of re-breathing and supported a hypothesis that NHF reduces 226 dead space. The clearance study was conducted during breath-holding. The effects of respiration on 235 clearance were excluded in this research to avoid the effect of breathing and due to the 236 technical restrictions. In several experiments there was no 81m Kr-gas clearance below the 237 soft palate (see also Figure 2B). This could be induced voluntarily, since it has been shown 238 that subjects can close their soft palate unintentionally during the breath-holding, but the 239 mechanism of this reflex is not fully understood (10). 240

241
Clearance of 81m Kr-gas in the lower parts of conducting airways may be of lesser relevance 242 due to very long half-times, as revealed; however, the fact that NHF can produce some 243 clearance even in those deep compartments may suggest a potential increase of the NHF 244 clearance efficiency with a presence of long end-expiratory pauses or opening of the mouth.  Inspired CO 2 is presented in Figures 3A and 4A as a total rather than as the first 100 mL per 279 breath, as with O 2 , because of high clinical relevance. 280

281
The ratio of CO 2 in the first 100 mL of inspired air to the total inspired CO 2 , as shown in 282 Figure 4D, resulted in a significantly higher ratio during NHF relative to the baseline (ratio = 283 0.84 (SD 0.10) during NHF vs. 0.75 (SD 0.12) at baseline; p < 0.01, paired t-test). This can be 284 explained by the clearance of expired gas in the upper airways that causes a reduction of the 285 last portion of re-inspired CO 2 measured in the trachea, thereby enhancing the ratio. 286 Therefore, when applying NHF, re-inspired CO 2 primarily results from the first 100 mL of the 287 inspired air, making the difference between the volumes of inspired CO 2 smaller and shifting 288 the ratio closer to 1.00. It can also be illustrated in Figure 3A, which shows most of CO 2 289 during NHF is measured within the first 100 mL and consequently increasing the ratio of CO 2 290 measured in 100 mL to CO 2 measured in the total inspired gas volume. The method of the 291 ratio calculation can be recommended for future studies as it is informative and may be used 292 without calibration of inspired volume. 293

294
Data on ventilation during the study (Table 3) shows a rather small amount of tidal volume 295 measured with RIP in all three patients. RIP was calibrated with a pneumotachograph before 296 and after the experiment and showed very small drift between calibrations, confirming the 297 robustness of the data. Nevertheless, tidal volumes smaller than 250 to 300 mL with normal 298 respiratory rate may suggest some inaccuracy of the method, which could affect volumes of 299 calculated inspired O 2 and CO 2 and lead to an underestimation of the parameters. It is 300 interesting to note that in two experiments minute ventilation was markedly reduced during 301 NHF while the respiratory rate was within normal values (range 10.6 to 15.0 min -1 ) and there 302 was no change in blood gases. Reduction of minute ventilation through a decrease of tidal 303 volume may indicate a reduction in the work of breathing without a change in blood gases, 304 which could remain clinically undetected because tidal volume is not measured routinely 305 during NHF therapy. Variability in the ventilation parameters shows that the effect of NHF on 306 ventilation in patients has to be investigated in the homogenous groups. The presence of a 307 probe in the trachea may also affect the breathing pattern and is preferably to be excluded 308 in such studies. and O 2 in the trachea during respiration confirmed the NHF-dependent decrease of re-319 breathing of expired air, which is eventually a reduction of dead space. 320

321
The reduction of dead space by NHF may increase alveolar volume if tidal volume remains 322 the same. It may also slow down the respiratory rate or reduce tidal volume and minute 323 ventilation, as has been observed in this study and also as previously reported in healthy 324 subjects during sleep (19). Reduction of the respiratory rate is the most frequently described 325 respiratory parameter associated with NHF therapy in adults and children (1, 16,26) and it is 326 also reported to be a simple and informative predictor of potentially serious clinical events 327 (3). It might be speculated that the reduction of respiratory rate by NHF can be more 328 substantial in patients with an increased respiratory rate. In this study the authors observed 329 very small reduction of the respiratory rate, which was within normal limits, but the small 330 sample size and the study design did not allow for any definitive conclusion. Reduction of 331 dead space may also affect gas exchange: a reduction of arterial CO 2 (1) , (20) and an increase 332 of oxygenation (7, 20) by NHF were shown, although these effects were not evident in this 333 study, probably, because the NHF application times (10 min) were too short. 334

335
The ratio of dead space to tidal volume increases during shallow breathing or when the total 336 physiological dead space is raised due to an increase of alveolar dead space in conditions like 337 emphysema, pulmonary embolism or ARDS (9, 13); this requires an increase of breathing 338 frequency to maintain the same level of alveolar ventilation. For the above-mentioned 339 conditions a small reduction of dead space would lead to a significant improvement in gas 340 exchange resulting in the reduction of minute ventilation, which would normalize blood gas 341 parameters or both.

Strengths and limitations 356
There are two key strengths in this current study. The first is the evaluation of dead-space 357 clearance without a breathing component, which is also a limitation and is outlined below. 358 The level of clearance is most efficient in the nasal cavities but may extend below the soft 359 palate; however, this has to be interpreted with caution. The data adds weight to the 360 argument that the respiratory support effects of NHF treatment are dependent not only on 361 the NHF rate but also on time; the longer the time during which NHF produces clearance at 362 the end of expiration, the more significant clearance can be expected. The second key 363 strength of the study is that the reduction of re-breathing by NHF was shown via a change of 364 actual gas composition in the inspired air. A correlation between the change of inspired 365 volumes of CO 2 and O 2 confirms the validity of the measurements. Elimination of CO 2 is of 366 primary interest, as a fraction of removed CO 2 from the expired gas is relatively higher than 367 the added fraction of O 2 and it is clinically relevant in hypercapnic patients. A role of 368 additional O 2 as a result of dead-space clearance in normo-and hypoxemic patients is yet to 369 be determined. 370

371
There are limitations to this study, however. The main drawback is that only static clearance 372 rates in the absence of breathing were quantified in the scintigraphy part. There were three 373 reasons to justify the design. First, 81m Kr-gas has a short lifetime (13 s) and it is a technical 374 restriction to visualize a fast-decaying radioactive tracer gas. Second, tidal breathing would 375 not allow studying the maximum clearance that can be potentially achieved by NHF. 376 Excluded in this study were investigations into the NHF clearance effects during a range of 377 tidal volumes, breathing patterns, opening the mouth, position of the soft palate, vocal 378 cords and the effects of changing the nasal prong size and position; these factors need to be 379 addressed separately in future study designs. Had the authors endeavored to include some 380 of these elements in the current study, they would have had to complicate the protocol 381 significantly and increase the number of patients in the group substantially, who would also 382 have needed to be homogeneous to allow adequate quantifications of individual responses. 383 The study of three tracheotomized patients was sufficient to demonstrate the NHF-384 dependent reduction of re-breathing as a physical process -although a large sample size in a 385 controlled trial would be required for the analysis of the above-mentioned parameters, 386 physiological responses or clinical outcomes of NHF therapy, which need to be studied 387 separately. It is unlikely that an increase of a sample size in the study without a change of 388 the design would lead to a valid conclusion on the physiological and clinical effects of NHF 389 therapy as the effects will greatly depend on the baseline parameters and duration of the 390 therapy. Frequent change of NHF rates during a relatively short time is not a desirable study 391 design for assessment of awake, spontaneously-breathing patients where an individual 392 voluntary response may affect the results. Also, a maximum NHF rate of 45 L/min was used 393 in this study in order to repeat the same three flows investigated in a model study (18) and 394 to limit the maximum radioactive daily exposure for the volunteers. In tracheotomized 395 patients there was a risk of non-completion of the protocol should another NHF rate be 396 added. Apart from the above, the authors could not exclude the fact that some patients 397   represented in the graphs, where the three symbols represent the three NHF rates applied. 583 The data in this figure is presented as means calculated from 2-minute intervals. An increase 584 of NHF from 15 to 45 L/min led to a flow-dependent reduction of inspired CO 2 and a rise of 585 inspired O 2 . C) Relation between change (Δ) of total inspired O 2 vs. CO 2 in the first 100 mL 586 per breath with linear regression (r 2 = 0.59) and 95% confidence intervals. This figure  587 demonstrates that there is a significant correlation between the reduction of CO 2 and the 588 increase of O 2 by means of NHF therapy (cc = -0.767, p = 0.016). D) Ratio of inspired CO 2 in 589 the first 100 mL of tidal volume to the total inspired CO 2 per breath during baseline 590 ventilation and during NHF (15, 30 and 45 L/min; ratio = 0.84 (SD 0.10) vs. 0.75 (SD 0.12) for 591 baseline measurements; p < 0.01). 592 593 594