key: cord-0260203-1lx615eq
authors: Boyle, K. G.; Eichenberger, P. A.; Schoen, P.; Spengler, C. M.
title: Comparing phrenic nerve stimulation using three rapid coils: implications for mechanical ventilation
date: 2022-04-05
journal: nan
DOI: 10.1101/2022.03.29.22272862
sha: 36ee91a603e8bf67aa9a25933c73186da9fd7dc3
doc_id: 260203
cord_uid: 1lx615eq

Rationale: Rapid magnetic stimulation (RMS) of the phrenic nerves may serve to attenuate diaphragm atrophy during mechanical ventilation. With different coil shapes and stimulation location, inspiratory responses and side-effects may differ. Objective: To compare the inspiratory and sensory responses of three different RMS-coils either used bilaterally on the neck or on the chest, and to determine if ventilation over 10min can be achieved without muscle fatigue and coils overheating. Methods: Healthy participants underwent 1-s RMS on the neck (RMSBAMPS) (n=14) with three different pairs of magnetic coils (parabolic, D-shape, butterfly) at 15, 20, 25 and 30Hz stimulator-frequency and 20% stimulator-output with +10% increments. The D-shape coil with individual optimal stimulation settings was then used to ventilate participants (n=11) for up to 10min. Anterior RMS on the chest (RMSaMS) (n=8) was conducted on an optional visit. Airflow was assessed via pneumotach and transdiaphragmatic pressure via esophageal and gastric balloon catheters. Perception of air hunger, discomfort, pain and paresthesia were measured with a numerical scale. Main results: Inspiration was induced via RMSBAMPS in 86% of participants with all coils and via RMSaMS in only one participant with the parabolic coil. All coils produced similar inspiratory and sensory responses during RMSBAMPS with the butterfly coil needing higher stimulator-output, which resulted in significantly larger discomfort ratings at maximal inspiratory responses. Ten of 11 participants achieved 10min of ventilation without decreases in minute ventilation (15.7{+/-}4.6L/min). Conclusions: RMSBAMPS was more effective than RMSaMS, and could temporarily ventilate humans seemingly without development of muscular fatigue.

The use of mechanical ventilation (MV) to replace spontaneous breathing is the gold standard during respiratory failure. However, the unphysiological positive pressure and prolonged diaphragmatic inactivity during MV can induce lung injury and diaphragm atrophy within 18h (1) . Atrophy is associated with intramuscular changes at fiber and cellular levels (2) (3) (4) , leading to a reduced ability to generate force and thus prolonged weaning time (5) , increasing the risk of further damage and pulmonary complications.

One potential method to reduce ventilator-associated diaphragm atrophy is to activate the diaphragm via nerve-stimulation. Preliminary results in animals (6, 7) have shown that phrenic nerve stimulation (PNS) reduces MV-induced diaphragm atrophy.

Furthermore, diaphragm pacing via PNS in spinal cord injured subjects can replace MV to induce resting breathing and preserve diaphragm function (8) . Intravenously inserted stimulation catheters have also been tested (9) and were recently shown to attenuate the decrease in inspiratory muscle strength, but to not decrease weaning time in difficult-to-wean patients (10) . However, any catheter poses a risk for infection and thus, non-invasive external PNS may serve as a favorable alternative.

Magnetic stimulation is one method for non-invasive PNS via a variety of coils and stimulation locations such as over the cervical spine (CMS) (11) , bilaterally and anterolaterally on the neck (BAMPS) (12) , or anteriorly on the chest (aMS) (13) . Two research groups have explored the use of rapid magnetic stimulation (RMS) in the context of diaphragm stimulation (14, 15) , and this technique was recently shown to not negatively interact with intensive care unit (ICU) equipment (16) . RMS may therefore serve as an ideal candidate for non-invasive treatment. Sander et al. (14) showed that bilateral anterior RMS (RMSBAMPS) can produce strong enough diaphragmatic contractions to induce short-term ventilation in healthy subjects for no longer than 5min due to "technical prerequisites." Adler et al. (15) sought to determine the optimal combination of stimulation-frequency and stimulator-output for tolerable rapid CMS (RMSCMS) in healthy humans, but they were unable to induce inspiratory flow despite sufficient diaphragm contraction.

With new developments of coils and cooling techniques, the present study aimed to compare inspiratory and sensory responses to RMSBAMPS when using different combinations of stimulation-frequency and stimulator-output with three differently-shaped coils, and to investigate whether a continuous series of RMSBAMPS could sustain ventilation at a constant level for 10min. Additionally, rapid anterior stimulation on the chest (RMSaMS) was tested, as this location may offer better nerve access.

The study was approved by the Cantonal Ethics Committee of Zurich (Project ID 2019-01990) and registered on clinicaltrials.gov (NCT04176744). The study conformed with the Declaration of Helsinki.

The study took place over 2-3 study visits. Participants abstained from intense exercise 48h before, and from any type of exercise 24h before each study visit, and from consumption of alcohol and caffeinated food or drinks on study days prior to testing. During visit 1, participants underwent lung function and respiratory muscle testing followed by a single-train RMSBAMPS protocol consisting of 1-s RMS trains of . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ;  https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint the phrenic nerves. The single-train RMSBAMPS protocol was conducted with three commercially available coils using various combinations of stimulator-output (% of maximum) and stimulation-frequency. During visit 2, participants had their body composition assessed prior to undergoing a series of consecutive trains of RMSBAMPS to induce ventilation in 1-min blocks with one of the three tested coils from visit 1 for a maximum of 10min. During the optional visit 3, participants underwent the same protocol as during visit 1, except that RMSBAMPS was replaced by RMSaMS performed on the chest. Cardiorespiratory variables, respiratory muscle activation, subjective perceptions and measures of side-effects were monitored throughout all visits.

Of the 15 participants originally recruited for the study, one participant withdrew due to an initial misunderstanding of study requirements. Thus, data of 14 participants (9M:5F) are presented. Due to the Covid-19 pandemic, only five participants (4M:1F) returned for the optional study visit and three additional participants (3F) were recruited at the end of the study to only undergo the optional visit 3 (RMSaMS).

Participant characteristics including anthropometrics, body composition, lung function and respiratory muscle strength are given in Table 1 .

In all participants standard lung function, respiratory muscle strength, and body composition was assessed. Lung function tests were performed using a commercially available testing system and body box (Quark PFT & Q-Box, Cosmed, Rome, Italy) according to current guidelines (17) . Respiratory muscle strength was assessed via maximal inspiratory and expiratory maneuvers (from residual volume and total lung capacity, respectively) alternating every three maneuvers using a respiratory pressure . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 5, 2022 to record transdiaphragmatic pressure (Pdi). Briefly, both balloon catheters were inserted through the nares, following numbing with local anesthetic spray (Xylocain Spray 10%, Aspen Pharma Schweiz GmbH, Baar, Switzerland), and both balloons were first placed in the stomach. One balloon remained in the stomach to measure gastric pressure (Pga). The second balloon was withdrawn in 1cm increments until a negative deflection was detected during a sniff maneuver, followed by the balloon being withdrawn an additional 10cm to ensure it was completely removed from the stomach in order to measure esophageal pressure (Pes). Participants were instructed to execute a Valsalva maneuver to empty both the gastric and esophageal balloons which were then filled with 2ml and 1ml of air, respectively. Final placement was . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint adjusted, if needed, so that end-expiratory pressure during tidal breathing resulted in a Pes of 0cmH2O. Both balloons were then secured to the nose with tape to ensure the same position throughout the experiments. Pdi was calculated as the difference between Pga and Pes. All participants elected to attempt catheter insertion but only 8 of the 14 tolerated the procedure. To determine if less invasive measures of diaphragm contraction would correlate with VT and Pdi, participants were also equipped with two respiratory belts (TN1132/ST, ADInstruments, Dunedin, New Zealand). One belt was placed over the naval, while one was placed along the nipple line in men and the highest possible position below the breasts in women to measure changes in abdominal and chest circumference, respectively (Abdominal and Chest). Finally, cardiovascular changes in response to RMS were monitored with a simple 3-lead electrocardiogram connected to a bioamplifier (PowerLab 15T, Dunedin, New Zealand) and a plethysmographic finger cuff (Nexfin, Edwards Lifesciences, Amsterdam, Netherlands) for continuous measurement of heart rate (HR) and blood pressure (BP), respectively.

Participants rated their perception of pain, discomfort and paresthesia in response to each RMS setting by pointing to a visual scale ranging from 0-to-10points. 0 was anchored as "none" while 10 was anchored as "maximal" (the maximum that one could imagine). Participants were also asked specifically whether they felt dental pain. During the continuous RMSBAMPS-ventilation trial on visit 2, the perception of air hunger was also evaluated via the same 0-to-10point visual scale. Change in galvanic skin response (GSR), a surrogate measure for changes in one's emotional response, was measured on the fingers using a commercially available GSR system (MLT118F . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. and D-shaped coils were also equipped with an active cooling unit (Coil cooler unit + high performance option, MagVenture, Farum, Denmark). All RMSBAMPS took place with the participant laying in a hospital bed that was tilted upright so that the torso was raised 30. Participants' heads were positioned within a vacuum cushion (Vacuform 2.0 vacuum pillow 30x40cm, Synmedic AG, Zurich, Switzerland) so that their necks were slightly extended. Positions of anatomical landmarks were recorded to ensure that all stimulations occurred with the participant in the same head and body position, as well as to ensure that the body position could be accurately repositioned on subsequent visits (for details also see below).

The single-train RMSBAMPS protocol began with 3min of resting breathing to quantify baseline measures and subsequently the placement of the first two coil anterolaterally on the neck in the position that yielded the highest response to stimulation. The coils were initially placed in the position that yielded the highest Pdi in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. at least twice unless participants indicated they could not tolerate a second train at that setting. In the event of data contamination (ie. swallow, upper airway collapse, improper timing, etc.), additional trains of RMSBAMPS were conducted at that setting on tester's discretion. Following RMSBAMPS with the four tested frequencies and all tolerable stimulator-outputs, the protocol was repeated starting with coil placement of the second pair of coils and after completion of that protocol, with the third pair. Coils were tested in a randomized order. Lastly, if a participant did not show any flow response, visit 1 was repeated on another day, at least 24h apart of the first visit 1. If . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint no flow could be initiated on the second day, these participants were considered as non-responders.

Participants in whom flow was induced via RMSBAMPS underwent thorough familiarization with continuous RMSBAMPS-ventilation. During this familiarization, the optimal stimulation-frequency and stimulator-output combination of visit 1 -using the pair of cooled coils that had yielded the largest VT at a submaximal stimulator-output -was tested at a respiratory frequency selected initially to match resting ventilation. If needed, adjustments to stimulator-output (5% increments), stimulation duration (+1ms increments) and respiratory frequency (+1breath-per-min increments) were made in attempt to optimally balance subjectively and objectively sufficient ventilation as well as suppression of participants' natural drive to breathe, while keeping their perception of air hunger, pain, discomfort and paresthesia tolerable throughout. As such, participants' verbal feedback was sought and taken into account to optimize their ability to complete the ten 1-min blocks of continuous RMSBAMPS-ventilation, as well as to reduce their urge for spontaneous breathing.

After familiarization, a 3-min resting breathing period was recorded followed by 1min of ventilation. Between each train of RMSBAMPS, participants passively expired and were instructed to not initiate the next inspiration. The flow signal was continuously monitored to ensure participants did not initiate breaths themselves. After each 1-min stimulation block, participants were asked to rate their perception of air hunger, pain, discomfort, and paresthesia, and whether they could tolerate a further minute of continuous RMSBAMPS. This break lasted the minimum amount of time to collect the . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint sensory ratings. This procedure was repeated until ten 1-min blocks were completed or until participant cessation.

The optional study visit 3 replicated the RMSBAMPS protocol from visit 1 (n=5)

with RMSaMS conducted on the chest. Coils were tested in two positions: 1) two coils placed bilaterally on either side of the chest; and 2) a single coil placed in the middle of the chest in attempt to stimulate both phrenic nerves simultaneously. A maximum of 1h was spent to optimize the position. The 3 additionally recruited subjects performed a DXA scan, lung function and respiratory muscle strength tests and the same RMSaMS protocol.

All physiological measurements were converted from analogue to digital with two 16-channel data acquisition systems (PowerLab 16/35, ADInstruments, Dunedin, New Zealand) and collected using LabChart Software (Version 8, ADInstruments, Dunedin, New Zealand) with a sampling frequency of 2,000Hz. VT was calculated by taking the integral of the flow measurement which was then BTPS corrected. Within a data analysis window that started at the beginning of RMS and ended at the absence of inspiratory flow ( Figure 1A) , VT, Pdi,mean and Pdi,peak, as well as mean Abdominal and Chest were calculated. Peak GSR was analyzed outside of this window due to the latency of this signal. All responses with the same RMS setting that did not induce total upper airway collapse and where pairs of stimulations were available, were included in the analysis of inspiratory variables and averaged. Stimulations that induced total upper airway collapse were not included in the analysis of inspiratory . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. variables, but were still quantified to determine the prevalence of this side-effect.

Responses during the continuous RMSBAMPS-ventilation trial were analyzed breath by breath in the same manner as the single-train RMS and averaged over 1 min, not including the short breaks. In addition, an average of all 1-min blocks was calculated to get a total overview of the RMSBAMPS-ventilation trial. The tension-time-index of the diaphragm (TTIdi,mean) was calculated as the average pressure the diaphragm produced during inspiration compared to the maximum a participant could achieve during a maximal maneuver (Pdi,mean/Pdi,max) multiplied by the duty cycle (inspiratory duration/breath cycle). Non-responders were not included in the analysis.

A number of factors contributed to an unequal sample size within and between coils at select RMS settings ( Figure 3A ) and between visits ( Figure 2) . First, not all participants were able to tolerate the same maximal RMS settings within and between coils. Second, not all participants were equipped with balloon catheters during each visit. Third, one participant was not able to undergo single-train RMSBAMPS with the butterfly coil due to too much facial paresthesia. Fourth, one participant was unable to return for visit 2 due to the COVID-19 pandemic. Finally, only five of 12 participants returned for the optional visit 3, so three additional participants were recruited to undergo visit 3 only. Therefore, the number of people included in the analysis is summarized in Figure 2 .

Due to the unequal sample size with each RMS setting, a mixed effects model two-way ANOVA with repeated measures was used to determine the effect of stimulation-frequency and stimulator-output on VT, Pdi, sensory ratings and GSR within each coil. Coils were compared between each other at select VT using a mixed . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Inspiration in response to single-train RMSBAMPS was induced in 12 of 14 participants. Both non-responders were female (with Pdi measurement) and

showed the same non-response on two visits. In general, increasing both stimulatoroutput and stimulation-frequency resulted in an increase in VT (n=12) and Pdi (n=6) as displayed in Figure 3B&D , with larger changes with stimulator-output compared to frequency increases. Increased stimulator-output had a significant effect on VT with all coils (all P0.003); while an effect of frequency was only present with parabolic . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint (P=0.023) and D-shape (P=0.041) coils, without an interaction effect. There was a significant effect of stimulator-output on Pdi,mean (all P0.005) and Pdi,peak (all P0.002) with all coils, while only the D-shape and parabolic coils had an effect of frequency with both variables (all P0.036), and no coil displayed an interaction effect. Both VT and Pdi,mean showed a stronger correlation with Abdominal than with Chest (Table 3) .

In general, sensory responses increased more prominently with increases in stimulator-output than increases in frequency ( Figure 3E -G). The largest stimulatoroutput tolerated in all stimulation-frequency and coil combinations was 20% ( Figure   3A ). For pain perception, all coils showed significant effects of stimulator-output (all For paresthesia, all coils had an effect of stimulator-output (all P0.004), but an effect of frequency was only present with parabolic (P=0.011) and butterfly (P=0.005) coils, without interaction with any coil. One participant was unable to undergo RMS with the butterfly coil citing too much facial paresthesia.

Changes in GSR in response to single-train RMSBAMPS are given in Figure 3H . is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint P<0.001), and a weak positive correlation with paresthesia (r=0.38, P<0.001) ( Table   3 ).

All incidents of UAC and dental pain are shown in Table 2 . UAC was a common side-effect most prevalent when stimulating at the lowest frequency (15Hz) and stimulator-output (20%). UAC was most likely to occur with the D-shape coil and least likely with the butterfly. The prevalence of dental pain (most prevalent with the butterfly coil) increased with increasing stimulator-output and occurred more often at 25 and 30Hz compared to 15 and 20Hz.

For any given stimulation-frequency and stimulator-output combination, the parabolic and D-shape coils achieved larger VT compared to the butterfly. With the butterfly, participants tolerated higher levels of stimulator-output achieving the same maximal VT as with the other coils at the highest tolerated stimulator-output (all P>0.05), but with significantly larger pain (vs. parabolic +1.21.7points, P=0.046; vs. D-shape +1.51.9points, P=0.030) at 30Hz. The highest achieved VT and the associated sensory ratings are presented in Figure 4 . Maximal VT did not differ between coils (parabolic 1.080.40L; D-shape 1.110.34L; butterfly 1.060.41L, P=0.804), but D-shape and butterfly coils needed more stimulator-output to achieve these volumes (parabolic=299%; D-shape=378%; butterfly=459%).

Corresponding Pdi,mean in participants who were equipped with balloon catheters during these stimulations was 10.75.5 (parabolic), 14.49.7 (D-shape) and

8.85.3cmH2O (butterfly). During these maximal stimulations, there was a significant effect of coil on discomfort (P=0.047) with post-hoc tests revealing significantly higher ratings between parabolic and butterfly coils (4.92.0 vs. 6.52.3points, P=0.046), but . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. 

Eleven participants underwent the RMSBAMPS-ventilation trial ( Figure 6 ) with the D-shape coil that did not overheat. Nine completed the 10-min protocol, one completed 6min (stopped due to discomfort; 7points), while in one participant ventilation could not be achieved. Five participants were hyperventilated compared to resting breathing . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint and a TTIdi,mean of 0.020.01 throughout the 10min, which did not change over time (P=0.833 and 0.937, respectively).

Eight participants underwent the single-train RMSaMS-protocol and only one participant (without balloons) showed a VT response, which could only be achieved with the parabolic coil. Inspiration was achieved with all stimulation-frequencies, but required ≥40% of stimulator-output. The maximal tolerated stimulator-output was 50% (15, 20, 25Hz ) and 40% (30Hz). The range of VT was 0.11-0.30L, while pain ranged between 5-7points, discomfort 7-8points and paresthesia 6-10points.

In the present study, RMSBAMPS could elicit flow in ≈86% of participants with parabolic, D-shape and butterfly coils, and the resultant diaphragm contractions induced VT similar or larger than during spontaneous resting breathing. The butterfly coil required the highest stimulator-output for VT similar to the other coils which resulted in larger discomfort at maximal responses. In contrast, bilateral RMSaMS could only elicit flow with the parabolic coil in one participant (13%). Finally, all but one participant (91%) reached 10min of continuous RMSBAMPS-induced ventilation, without the coil overheating or a decrease in V E over time.

While all three coils produced similar inspiratory responses, the butterfly coil required the largest stimulator-output despite having the largest maximal magnetic flux. The discrepancy between responses of different coils at the same stimulator-. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint output likely reflects the butterfly being the largest coil making it difficult to optimally place often resulting in a small space between skin and coil surface. The higher stimulator-output needed to produce similar VT also resulted in the largest discomfort at maximal responses, possibly due to the butterfly coil causing more dental pain.

Adler and co-workers (15) reported that 15Hz at 65% of stimulator-output served as the optimal setting using RMSCMS to achieve the highest Pdi,peak (≈20 cmH2O) with sensory ratings below 3points. Using the same criteria in the present study (pain below 3points; 4 participants equipped with balloons), the optimal settings would be 25Hz at 30% stimulator-output with the parabolic (Pdi,peak=14.84.5cmH2O), 20Hz at 40% with the D-shape (Pdi,peak=19.65.6cmH2O), and 15Hz at 40% with the butterfly (Pdi,peak=8.42.9cmH2O) coils. These Pdi, as well as maximal Pdi, are all lower than the ones of Adler et al. (15) . However, this difference likely reflects that our participants kept their glottis open to allow airflow, while Adler et al.'s (15) kept their glottis closed during stimulation. Also, CMS is known to produce larger Pdi twitches in response to single stimuli compared to bilateral stimulations (13, 21) , attributed to recruitment of accessory respiratory muscles leading to chest wall stiffening (21) , that likely also applies during RMSCMS. The diaphragm activation achieved in the present study in response to RMSBAMPS reached and exceeded normal resting VT, and a level of stimulation likely able to attenuate diaphragm atrophy during MV. Recently, Sotak et al. (22) showed that percutaneous electrical PNS in ICUpatients during MV keeping the work of breathing between 0.2-2.0J/L reduced the rate of the development of diaphragmatic atrophy. However, in order to maximize therapeutic effects, methods to increase the level of activation without increasing VT and thus without increasing the risk for lung injury, should be further explored.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. Finally, a non-invasive abdominal belt may serve as a useful tool to quantify the amount of diaphragm activation during RMS given that Abdominal correlated with both VT and Pdi,mean. Similarly, given that GSR correlated with both pain and discomfort, it may serve as a useful measure of distress in unconscious patients (Table 3) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint phrenic nerve stimulation (23)) may contribute to this discrepancy. Notably, despite flow being induced in the majority of our participants, UAC was still prominent. All but one participant experienced full or partial UAC at least once during single-train RMSBAMPS and two participants appeared to experience total UAC during all singletrains over two visits despite adjustments in stimulator-output, stimulation-frequency and coil positions. However, UAC may be an irrelevant issue in intubated and noninvasively ventilated patients given that Adler et al. (15) successfully alleviated UAC during RMSCMS when positive pressure ventilation was added. Thus, positive airway pressure may be needed to avoid closure, but this needs to be explored more systematically.

Continuous RMSBAMPS induced ventilation in ten of 11 participants, while one participant experienced consistent UAC, despite responding to single-train RMSBAMPS.

The 10-min limit was a result of our protocol rather than equipment overheating or participant intolerance (all that achieved 10min indicated they could go longer). Thus, with proper coil cooling, it is possible to overcome the 5-min limitation seen by Sander et al. (14) .

Inducing ventilation via RMSBAMPS appears to be a tolerable technique given the majority (90%) of participants with continuous ventilation were able to complete the 10-min protocol. In fact, mean perception of pain, discomfort and paresthesia were rated below 3points, however, one participant ended the trial early citing discomfort resulting from excessive shoulder movement on one side. This resulted from the coil moving out of the optimal position during the trial and reflects the importance of maintaining the same position between consecutive stimulations. All but two . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint participants required slightly higher than 1-s stimulation duration (up to 1.3s) for optimal comfort, and six participants required a higher respiratory frequency compared to their resting breathing to suppress their urge to breathe. These adjustments occurred during familiarization based on their feedback and resulted in a slight hyperventilation in five cases. Perception of air hunger was rated ≤2 points at all times in all but two participants. It should be noted, however, that a brief break was taken between each minute to assess participant's sensory ratings in which they resumed spontaneous breathing which may have contributed to reduced sensory ratings. Thus, the efficacy of ventilation without breaks and a longer than 10min still needs further exploration.

The mean V E achieved in the present study was similar to the 14.0L/min by Sander et al. (14) with 25Hz at 40% stimulator-output (two MagStim butterfly coils) in 10 healthy volunteers. Their ventilation, however, surpassed the present when stimulatoroutput was increased to 50% (18.6L/min), but this increase in stimulator-output resulted in only three participants undergoing ventilation due to a largely increased perception of pain and discomfort, similar to our single-train RMSBAMPS responses with maximal stimulator-output. The present study optimized RMS settings to each participant and was 20 (n=3), 25 (n=5) or 30 (n=2) Hz with a mean stimulator-output of 27% (range=20-40%). Stimulation parameters and thus VT likely exceeded the levels that would be used or needed in mechanically ventilated patients in order to prevent lung injury. In any case, it is encouraging to note that in the present study with an excessive ventilation present, mean TTIdi was only 0.02 which is well below the 0.15 fatigue-inducing threshold (24) . As such, RMSBAMPS seems unlikely to cause fatigue-inducing contractions, at least in our group of young healthy participants, and . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint as previously shown with CMS (15) . To which extent muscle fatigue could play a role in various patient groups needs, however, further exploration.

Finally, while Sander et al. (14) reported no cardiovascular changes (unspecified), the present study showed an increased BP during RMSBAMPS-ventilation, but without changes in HR. Although the increase in BP did not reach an unsafe level in these young subjects and stimulator-output was likely higher than what would be used in patients, it is still suggested that the cardiovascular system is closely monitored during continuous RMS.

A few limitations remain. First, although flow traces were monitored to guarantee participants did not initiate inspiration, volitional assistance cannot be excluded with certainty. Second, the COVID-19 pandemic impacted the number of participants that returned for visit 2 and the optional visit. Third, not all participants tolerated balloon catheters resulting in a reduced sample with Pdi data, two being nonresponders.

RMSBAMPS, but not RMSaMS, can induce strong enough diaphragmatic contractions to ventilate healthy humans for 10min. However, we currently do not recommend replacing MV with RMS, but rather to use RMS to assist MV in order to potentially reduce ventilator-induced diaphragm atrophy. For most effective clinical use, newly designed equipment should be less bulky and optimized for PNS with stimulators automatically adjusting timing and output according to feedback from ventilation and non-desired side-effects.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. participants (five with balloons) could be analyzed with the butterfly coil. One participant (without balloons) was unable to return for the RMSBAMPS-ventilation trial . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint (visit 2) due to the COVID-19 pandemic, while one participant who was equipped with balloons on visit 1 declined to undergo balloon catheter insertion on visit 2. One additional participant (without balloons) was unable to be ventilated with continuous RMSBAMPS. As such, ten participants (five with balloons) could be analyzed for RMSBAMPS-ventilation. Five participants elected to return for the optional visit in which single-train RMS was conducted on the chest, and three additional participants were recruited to undergo this visit only. Only one participant (without balloons) showed an inspiratory response to RMS on the chest with the parabolic coil only and could be analyzed. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 5, 2022. (19) . Predicted maximal voluntary ventilation was obtained by multiplying FEV1 by 35. Secondary recruitment represents the participants who were recruited for the optional third visit only (rapid magnetic stimulation on the chest).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint Table 3 . Group and individual correlations for respiratory belts with inspiratory responses, and change in galvanic skin response with sensory responses during single-train rapid magnetic stimulation bilaterally on the neck.

Definition of abbreviations: Abdominal = change in abdominal circumference; VT = tidal volume; Pdi,mean = mean transdiaphragmatic pressure; Chest = change in chest circumference; GSR = change in galvanic skin response. Individual Pearson correlations were performed between respiratory belts and inspiratory variables, and individual Spearman correlations were performed between GSR and sensory responses. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 5, 2022. ; https://doi.org/10.1101/2022.03.29.22272862 doi: medRxiv preprint

Rapid disuse atrophy of diaphragm fibers in mechanically ventilated humans

Rapidly progressive diaphragmatic weakness and injury during mechanical ventilation in humans

Dysfunction of respiratory muscles in critically ill patients on the intensive care unit

Diaphragmatic dysfunction in patients with ICU-acquired weakness and its impact on extubation failure

Can phrenic stimulation protect the diaphragm from mechanical ventilation-induced damage?

Phrenic Nerve Stimulation Protects against Mechanical Ventilation-Induced Diaphragm Dysfunction in Rats

Phrenic nerve pacing via intramuscular diaphragm electrodes in tetraplegic subjects

Temporary transvenous diaphragm pacing vs. standard of care for weaning from mechanical ventilation: study protocol for a randomized trial

Investigators R-SG. Randomised Clinical Study of Temporary Transvenous Phrenic Nerve Stimulation in Difficult-to-Wean Patients

Cervical magnetic stimulation: a new painless method for bilateral phrenic nerve stimulation in conscious humans

Bilateral magnetic stimulation of the phrenic nerves from an anterolateral approach

Anterior magnetic phrenic nerve stimulation: laboratory and clinical evaluation

Electromagnetic Ventilation: First Evalution of a New Method for Artifical Ventilation in Humans

Repetitive magnetic stimulation of the phrenic nerves for diaphragm conditioning: a normative study of feasibility and optimal settings

Assessment of magnetic flux density properties of electromagnetic noninvasive phrenic nerve stimulations for environmental safety in an ICU environment

Standardisation of spirometry

Multi-ethnic reference values for spirometry for the 3-95-yr age range: the global lung function 2012 equations

Predicted normal values for maximal respiratory pressures in caucasian adults and children

of Intensive Care Medicine/Society of Critical Care Medicine Clinical Practice Guideline: Mechanical Ventilation in Adult Patients with Acute Respiratory Distress Syndrome

Cervical magnetic stimulation as a method to discriminate between diaphragm and rib cage muscle fatigue

Phrenic nerve stimulation prevents diaphragm atrophy in patients with respiratory failure on mechanical ventilation

Effects of neck flexion and mouth opening on inspiratory flow dynamics in awake humans

Effect of pressure and timing of contraction on human diaphragm fatigue

International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity