Diaphragm neurostimulation during mechanical ventilation reduces atelectasis and transpulmonary plateau pressure, preserving lung homogeneity and PaO2/FiO2.

Tidal volume delivered by mechanical ventilation to a sedated patient is distributed in a non-physiological pattern, causing atelectasis (underinflation) and overdistension (overinflation). Activation of the diaphragm during mechanical ventilation provides a way to reduce atelectasis and alveolar inhomogeneity, protecting the lungs from ventilator-induced lung injury while also protecting the diaphragm by preventing ventilator-induced diaphragm dysfunction. We studied the hypothesis that diaphragm contractions elicited by transvenous phrenic nerve stimulation, delivered in synchrony with volume-control ventilation, would reduce atelectasis and lung inhomogeneity in a healthy, normal-lung pig model. Twenty-five large pigs were ventilated for 50 hours with lung-protective volume-control ventilation combined with synchronous transvenous phrenic-nerve neurostimulation on every breath, or every second breath. This was compared to lung-protective ventilation alone. Lung mechanics and ventilation pressures were measured using esophageal pressure manometry and electrical impedance tomography. Alveolar homogeneity was measured using alveolar chord length of preserved lung tissue. Lung injury was measured using inflammatory cytokine concentration in bronchoalveolar lavage fluid and serum. We found that diaphragm neurostimulation on every breath preserved PaO2/FiO2 and significantly reduced the loss of end-expiratory lung volume after 50 hours of mechanical ventilation. Neurostimulation on every breath reduced plateau and driving pressures, improved both static and dynamic compliance and resulted in less alveolar inhomogeneity. These findings support that temporary transvenous diaphragm neurostimulation during volume-controlled, lung-protective ventilation may offer a potential method to provide both lung- and diaphragm-protective ventilation.

strategy that protects both the diaphragm and the lungs at the same time, but it is not self-evident that the benefits of diaphragm contraction to preclinical subjects that are deeply sedated, thereby offering a 110 potential mechanism to prevent the development of all the aspects that contribute to VILI and VIDD 111 during the acute sedation phase of their ICU course in humans. Our primary hypothesis is that bronchoscopes due to the length of the trachea in pigs of this size. At necropsy, lung tissue samples were  ensure that groups were large enough to be able to use nonparametric statistical tests. Additional 168 animals were added to the MV, MV+TTDN50% and MV+TTDN100% groups to replace lost tissue samples 169 from a freezer failure, procedural challenges during the 2020 COVID-19 pandemic, and as part of an 170 effort to balance experimental pairs. Statistical analysis was performed using GraphPad Prism version 171 8.0.0 (GraphPad Software, San Diego, California USA, www.graphpad.com). All measurements are reported as median (IQR), 95% confidence interval of the median and were tested using nonparametric There was a difference between subject weights in all groups in the analysis of variance, with 186 MV+TTDN100% being larger than the MV+TTDN50% group as per the post-hoc analysis, however the 187 length-to-weight ratio was the same for all four groups (Table 1). There was no difference in fluid 188 balance, anesthetic drug dose, ventilator settings or blood gas values between ventilated groups (Table   189 2). Controlled mechanical ventilation resulting in normal arterial blood gas values was achieved in all 190 ventilated groups, for the duration of the 50-hours. The mixed-effects analysis identified that PaO 2 and 191 PaCO 2 were both affected by time but not by TTDN dose or the interaction of TTDN dose with time.    Figure 1A). Tidal volume in the ventral lung region was also 53) for breaths with no diaphragm contraction (passive) to 53% (50-57) for breaths when the diaphragm 205 was contracted (active) in the same animal of the MV+TTDN50% group, with a concurrent shift in ventral 206 tidal volume, 53% (47-55) to 47% (44-51) in the same animal of the MV+TTDN50% group, p=0.0499. Figure 1B shows an example of an EIT image documenting the redistribution of tidal volume from ventral 208 to dorsal lung regions between passive and active breaths in the same animal.

232
The alveolar chord length is different between all groups for each tissue sample location (p<0.0001 233 location 1, p=0.0045 location 2, p<0.0001 location 3, p<0.0001 location 4, p<0.0001 location 5). Samples group in the post-hoc analysis. Refer to Table 5. Figure 3 shows hematoxylin and eosin-stained sections 236 of lung tissue from each sample location and group.

246
Esophageal pressure measured at end-inspiration was not different at baseline or study end, between 247 groups or for the interaction of TTDN dose with time. Esophageal pressure was affected by time, 248 p=0.0214 (Table 6). Plateau pressure was different between the three groups at study end, p=0.0258.

249
The mixed-effect analysis shows that plateau pressure was affected by time (p=0.0054) and TTDN dose 250 (p=0.0203). The interaction of TTDN dose with time was not significant. Driving pressure (plateau 251 pressure minus PEEP) was different at study end between the three groups, p=0.0258. The mixed-effect 252 analysis shows that driving pressure was affected by time (p=0.0118) and TTDN dose (p=0.0251). The    and study end (p<0.0001). The mixed-effect analysis shows that dynamic compliance was not affected by 261 time but the effect of TTDN dose was significant (p<0.0001). The interaction of TTDN and time was also 262 significant, p=0.0048. The slope of the linear regression was different between all three groups significant. Measurements were taken for all breaths (MV or MV+TTDN as dictated by the study group).

272
The mixed-effect analysis shows that PaO 2 /FiO 2 was affected by time (p<0.0001) and TTDN dose 273 (p=0.0393). The interaction of TTDN dose and time was not significant ( Table 6)

280
Inflammatory cytokines in the BALF were the lowest in the MV+TTDN100% group but did not reach 281 statistical significance. There were also no statistical differences between inflammatory cytokine 282 concentrations measured in serum or in the modified lung injury scores performed on histology samples.

283
Refer to Table E2 in the Online Data Supplement for details.

286
No instances of ventilator asynchrony, such as double-triggered breaths or breath stacking, were 287 observed. We manually reviewed pressure and flow waveforms recorded at the termination of the 288 ventilator circuit, as well as esophageal pressure tracings, and no evidence of reverse triggering was We had two instances where mucus plugs were mobilized by diaphragm contractions that plugged-off an 291 area of the lung in the MV+TTDN50% group in the last 6-8 hours of the experiment. These episodes 292 resolved over 2-4 hours with suctioning but required a temporary increase in FiO 2 for a period. This did 293 not occur in the other groups; however, all groups were suctioned every 6 hours routinely to ensure that 294 any derecruitment due to suctioning was similar between groups. Extravascular lung water was not 295 significantly different between groups ( Figure

308
The significant mitigation of end-expiratory lung-volume loss with the use of adjunctive TTDN on every 309 breath is an important finding and establishes a biological foundation for the potential clinical benefit of 310 TTDN in critically ill patients. There was twice as much end-expiratory lung volume lost in the MV group 311 compared to the MV+TTDN100% group, with the MV+TTDN50% group being closer to the MV group.

312
This demonstrates that TTDN on every breath preserves more end-expiratory lung volume than it does 313 on every second breath, thus the dose of TTDN is important. The more alveoli that are preserved for tidal atelectrauma and cyclic strain. This is further supported by the improved PaO 2 /FiO 2 , lower 316 transpulmonary plateau pressure and respiratory system driving pressures in the MV+TTDN100% group 317 versus the MV+TTDN50% group. TTDN on every breath reduced driving pressure in comparison to the 318 gold-standard lung-protective ventilation strategy alone. We have previously shown that TTDN on every 319 second breath, targeting the same ventilator PTP reduction as this study, mitigates ventilator-induced

381
However, a TTDN contraction is completely synchronous as it follows the ventilator during inspiration 382 and the strength of the contraction is adjusted to reach a targeted PTP reduction. (29) This study 383 demonstrated no evidence of increased injury due to TTDN as lung injury scores were the same for all 384 groups and inflammatory markers were not significantly different for any group. Extravascular lung 385 water was also not significantly different, therefore TTDN did not cause an adverse transcapillary lung water such as is the case with negative-pressure pulmonary edema. Variation in tidal volume was 388 prevented by the use of volume-controlled breaths and there was no observed ventilator asynchrony or 389 reverse triggering. The lack of ventilator asynchrony and absence of increased signs of lung injury in the 390 TTDN groups support that TTDN does not increase the risk of lung injury and that it is reasonable to 391 investigate the potential benefits of TTDN in patients with acute lung injury.

393
Our study has certain limitations, as it uses a preclinical model with healthy lungs and only a small 394 number of subjects. We recognize that many ICU patients do not fit this model, but the patients who 395 require extended sedation and mechanical ventilation, and are the most at risk for VIDD, do fit this 396 model, and we have previously established that TTDN can offer protection from VIDD in healthy pigs.

397
(29) The goal of this study was to investigate the effects that TTDN can have on ventilator-induced lung 398 injury, the results of which may apply to heavily sedated, critically ill patients receiving controlled 399 mechanical ventilation. Whether the results are the same in human ICU subjects who are not healthy 400 remains to be seen. This study was limited in its ability to measure transpulmonary driving pressure and 401 transdiaphragmatic pressure as we did not obtain a full data set due to late equipment acquisition. While 402 pressure and compliance data are encouraging, more transpulmonary and transdiaphragmatic pressure 403 data are needed to be conclusive. Critically ill patients that are affected by VILI are typically ventilated for 404 more than 50 hours, however a longer duration of this experiment would have been technically difficult.

405
The ability to do caudal lobe bronchial washings was inhibited by the length of bronchoscopes available, 406 thus we could not fully investigate the level of lung inflammation and possible injury. Future 407 investigation would benefit from using an injured-lung model to adequately evaluate the effect of TTDN 408 on VILI in a nonhomogeneous, injured lung, to better reflect commonly encountered clinical scenarios.
Not only is TTDN beneficial but the manner in which it is delivered provides a specific benefit beyond

426
This study is important with findings that will inform the application of a promising new therapeutic 427 modality. Our findings show that TTDN on every breath preserved PaO 2 /FiO 2 and required lower driving 428 pressure than the reference-standard lung-protective mechanical ventilation strategy over 50 hours in 429 healthy lungs. There was less end-expiratory lung volume loss and better alveolar homogeneity with 430 TTDN on every breath. Furthermore, TTDN resulted in reduced atelectasis and the accompanying 431 potential for lung injury. This offers insights into the impact that atelectrauma has on VILI in non-injured 432 lungs and its associated alveolar inhomogeneity. This study supports the hypothesis that synchronous 433 diaphragm contraction on every breath during controlled mechanical ventilation is beneficial and is no exacerbate alveolar instability after surfactant deactivation. Crit.Care Med. 30: 12: 2675Med. 30: 12: -2683Med. 30: 12: , 2002 ventilation distribution in horizontal man. J. Appl.Physiol. 40: 3: 417-424, 1976.