Phrenic Nerve Block and Respiratory Effort in Pigs and Critically Ill Patients with Acute Lung Injury

These studies were approved by two different ethics committees: one for animal experiments and one for clinical studies. Both protocols were funded by the University of São Paulo Medical School (São Paulo, Brazil). For further information and complete methods, please refer to Supplemental Digital Content 1 (https://links.lww.com/ALN/C800).

After the approval by the Ethics Committee for Animal Studies (No. 967/2017, Faculdade de Medicina da Universidade de São Paulo), six female Landrace pigs with average weights of 36.9 kg were chosen (bladder catheterization in females was easier). The experimental protocol has been described previously.⁶  Briefly, pigs monitored with electrocardiogram and pulse oximetry were anesthetized with intramuscular ketamine and midazolam, followed by intravenous midazolam, fentanyl, and pancuronium (a neuromuscular blocking agent). An internal jugular vein catheter and femoral artery line were placed to sample blood gases, administer vasoactive drugs, and monitor blood pressure. An esophageal balloon (Nutrivent; Sidam, Italy) was inserted to allow measurement of P_eso; correct placement was verified using the occlusion technique.²¹  We measured electrical activity of the diaphragm using a dedicated Neurally Adjusted Ventilatory Assist catheter (Maquet Critical Care, Sweden). Electrical impedance tomography was recorded using an Enlight monitor (Timpel, Brazil) with 32 electrodes embedded in a customized silicon belt placed on the perimeter, defining a cross-sectional plane of the thorax at the level of the sixth intercostal space (i.e., parasternal line).

The experimental protocol consisted of two phases: (phase 1) lung injury and pressure support titration, and (phase 2) phrenic nerve blockade (fig. 1). In phase 1, we induced severe ARDS by lung lavage with Tween (an astringent solution) followed by injurious mechanical ventilation until a Pao₂/inspiratory oxygen fraction ratio less than or equal to 150 mmHg with a positive end-expiratory pressure (PEEP) of 10 cm H₂O (during the whole procedure) was reached under deep sedation (propofol, ketamine, and remifentanil). Then anesthesia was titrated down until spontaneous inspiratory effort reached a ΔP_eso greater than or equal to 10 cm H₂O during an airway occlusion procedure.⁶  After lung injury and sedation titration, 1 mg/kg succinylcholine and 1 to 2 mg/kg propofol were administered, and pigs were briefly ventilated on pressure-controlled ventilation. We calculated respiratory system compliance (C_dyn), and titrated inspiratory pressure to guarantee a V_T greater than 4 ml/kg. After succinylcholine and propofol had weaned and the pig was back to assisted ventilation, and regardless of the pig’s inspiratory effort, this guaranteed V_T greater than 4 ml/kg after bilateral phrenic nerve block. The inspiratory pressure was kept unchanged during the entire protocol.

In phase 2, we performed bilateral phrenic nerve block using ultrasound imaging and a neurostimulator to identify the cervical plexus by the motor response on the ipsilateral arm (Plexygon, Vygon, Italy). The identification was followed by the injection of 20 ml lidocaine, 2%. Please refer to Supplemental Digital Content 2 (https://links.lww.com/ALN/C828) for further information.

Electrical activity of the diaphragm, ΔP_eso, V_T, peak transpulmonary pressure, driving pressure, and dorsal and left lung ventilation were monitored. Peak transpulmonary pressure was calculated as pressure support − ΔP_eso during spontaneous breathing. Driving pressure was calculated as V_T/C_dyn because no inspiratory pauses were performed. Please refer to Supplemental Digital Content 3 (https://links.lww.com/ALN/C829) for further information. For ventilation distribution analysis, the electrical impedance tomography image was divided into the right and left lung, and into four regions of interest, each covering 25% of the ventrodorsal diameter. The percentage of ventilation in the most dependent region of the lung was calculated as reflecting the inspiratory activity of the diaphragm and being the most gravity-dependent zone. Please refer to Supplemental Digital Content 4 (https://links.lww.com/ALN/C830) for further information. The percentage of left lung ventilation was analyzed to estimate any changes in ventilation distribution that could have been induced by the sequential phrenic nerve blockade. During assisted ventilation, pendelluft was quantified, a phenomenon that occurs when a strong diaphragmatic contraction leads to deflation of the ventral portion of the lung and overdistention of the dorsal portion of the lung during early inflation²²  and that can cause regional lung injury.⁶  Data acquisition in pigs occurred at five time points: before phrenic nerve blockade during assisted ventilation (i.e., baseline); 4 min after left phrenic block; 10 min after the left phrenic block; 4 min after the right phrenic block; and 10 min after right phrenic block.

Our primary goal was to evaluate whether this technique was feasible and efficient at reducing V_T or peak transpulmonary pressure. Secondary objectives were to quantify the decrease in electrical activity of the diaphragm, ΔP_eso, and driving pressure, as well as changes in the distribution of ventilation.

After obtaining Institutional Review Board approval (Certificate for Ethics Presentation and Appreciation number: 02029118.2.0000.0068), patients admitted for ARDS and under assisted spontaneous breathing were screened for inclusion in the study between May 2019 and September 2020. Informed consent was obtained from legal guardians. Of note, written informed consent was obtained for the first three patients, but during the COVID-19 outbreak, strict visiting policies were implemented, and oral consent was authorized and taken from the substitute decision-maker via telephone. Data collection was carried out in the respiratory intensive care unit at Heart Institute (Instituto do Coração) of the Hospital das Clínicas (University of São Paulo Medical School). This study was registered at clinicaltrials.gov (NCT03978845).

Patients eligible for the study protocol had to be older than 18 yr; on invasive pressure support ventilation with a Pao₂/inspiratory oxygen fraction ratio less than 300 mmHg triggering the ventilator; and a driving pressure greater than 15 cm H₂O or a V_T greater than 10 ml/kg of predicted body weight. We evaluated if a reliable plateau pressure could be obtained and calculated the driving pressure by subtracting PEEP from plateau pressure during an inspiratory pause.²³  If driving pressure was greater than 15 cm H₂O and/or V_T was greater than 10 ml/kg, the patient fulfilled inclusion criteria. Exclusion criteria included use of long-acting neuromuscular blocking agents in the previous 3 h; pain or Richmond Agitation–Sedation Scale score greater than 0; arterial pH less than 7.25; hemodynamic instability or a need for increased doses of vasopressor drugs in the previous 2 h; intracranial hypertension; presence of chest or abdominal fistula; neuromuscular disease; spinal cord trauma; massive ascites; burns in the thoracic region; tetanus; and pregnancy.

The Richmond Agitation–Sedation Scale was used for measurement of their level of sedation; in addition, two patients were monitored via Bispectral Index to assess sedation depth. An electrode belt (Timpel Enlight 1800 Model; Timpel) was installed in the thoracic region between the fourth and fifth intercostal spaces. Two esophageal catheters (Nutrivent and Neurally Adjusted Ventilatory Assist catheter) were positioned and calibrated to measure esophageal pressure and electrical activity of the diaphragm, respectively. Flow, pressures (peak, plateau, PEEP), and V_T were continuously recorded from the electrical impedance tomography device and from the SERVO-i ventilator (Maquet Critical Care) connected to a laptop using the Servo Tracker application (Maquet Critical Care). ΔP_eso was monitored using either the Optivent (Optivent Sidam, Italy) or the Pneumodrive device (Bionica, Brazil). Correct placement of the esophageal catheter was verified using the occlusion technique.²¹  Heart rate, blood pressure, and peripheral oxygenation were continuously monitored. After inclusion of the patient in the protocol, sedation was not changed, and vasoactive drugs were titrated at the discretion of the attending physician.

The clinical protocol consisted of three phases: (phase 1) inspiratory pressure titration, (phase 2) bilateral phrenic nerve block, and (phase 3) recovery phase (fig. 2). During phase 1, succinylcholine (1 mg/kg) and propofol (1.5 to 2.5 mg/kg) were administered, and patients were briefly ventilated on pressure-controlled ventilation. We calculated respiratory system compliance, and titrated inspiratory pressure that was kept unchanged after the effect of succinylcholine and propofol had weaned. This way, regardless of the patient’s inspiratory effort, we would guarantee V_T greater than 4 ml/kg of after bilateral phrenic nerve block.

During phase 2, a trained anesthesiologist performed phrenic nerve block using an ultrasound (Mindray M6 Ultrasound System; China) and a peripheral nerve stimulator (Stimuplex HNS 12B Braun; Germany). Of note, an anesthesiologist who has performed 15 peripheral blocks is able to distinguish anatomical landmarks and has a success rate close to 90%.²⁴  Please refer to Supplemental Digital Content 5 (https://links.lww.com/ALN/C831) for further information. The combination of ultrasound and neurostimulator minimizes the risk of intravascular or intraneural injection, neural damage, and vascular bruising.²⁵  Use of ultrasound minimizes the risk of complications and allows observation of local anesthetic dispersion; the neurostimulator avoids intraneural administration of local anesthetic, thus minimizing risk of a potential association with neural damage.²⁵  When the patient has motor response at low current output (e.g., less than 0.2 mA), one cannot ensure that the needle is not intraneural. We did not administer lidocaine if the phrenic nerve could not be visualized or if the patient had a motor response at a current output that was lower than 0.50 mA, greatly minimizing risk of intraneural injection.²⁶  Finally, we used lidocaine due to its pharmacokinetic and pharmacodynamic properties: shorter half-life, better cardiovascular stability, low latency, and faster onset than other local anesthetics.²⁷ 

The study protocol was conducted under a predefined sequence: the protocol was initiated on the left nerve, then followed by blockade of the right phrenic nerve 10 min later. We assessed the phrenic nerve block with three tools: electrical activity of the diaphragm signal, ΔP_eso, and regional ventilation distribution on continuous positive airway pressure. We evaluated the (1) percentage of patients blocked, (2) amount of time spent from scanning to administration of lidocaine, and (3) duration until recovery. Full recovery from bilateral phrenic nerve blockade was defined as the time after bilateral nerve block when the electrical activity of the diaphragm, ΔP_eso, and V_T were within ±20% of baseline values.

Electrical activity of the diaphragm, ΔP_eso, V_T per predicted body weight, peak transpulmonary pressure, driving pressure, respiratory rate in breaths per minutes, and ventilation distribution were analyzed. Peak transpulmonary pressure was calculated as pressure support − ΔP_eso. Driving pressure was calculated as V_T/C_dyn because no inspiratory pauses were performed. Please refer to Supplemental Digital Content 3 (https://links.lww.com/ALN/C829) for further information. Ventilation distribution was also assessed. Finally, the percentage of left lung ventilation was analyzed. Please refer to Supplemental Digital Content 4 (https://links.lww.com/ALN/C830) for further information. Data acquisition occurred at seven time points: before phrenic nerve blockade during assisted ventilation (i.e., baseline), 4 min after left phrenic block, 10 min after the left phrenic block, 4 min after the right phrenic block, 10 min after right phrenic block, 1 h after right phrenic block, and on the following day (i.e., final).

The primary goal was to evaluate the reduction in V_T or peak transpulmonary pressure. Secondary objectives included changes in electrical activity of the diaphragm, ΔP_eso, driving pressure, and respiratory rate. We also analyzed ventilation distribution using electrical impedance tomography and assessed whether a bilateral phrenic nerve block could either reduce or abolish pendelluft. Finally, we aimed to evaluate the time to wean from the block, as well as feasibility, safety, and efficacy of this technique.

No statistical power calculation was used to guide sample size. An initial sample size of 10 patients was planned, but only 9 were included because a Neurally Adjusted Ventilatory Assist catheter malfunctioned. The unadjusted results are reported as median and interquartile range. We fitted linear mixed models for each outcome, accounting for the repeated measurements on each subject with a random intercept. Models were adjusted for time (baseline, 4 min after left phrenic block, 10 min after the left phrenic block, 4 min after the right phrenic block, 10 min after right phrenic block, 1 h after right phrenic block, and final), PEEP (continuous variable), and pressure support (continuous variable) at each time point for humans; and for time (baseline, 4 min after left phrenic block, 10 min after the left phrenic block, 4 min after the right phrenic block, 10 min after right phrenic block) and pressure support (continuous variable) at each time point for pigs. Data from pigs and humans were analyzed separately. No imputation was performed for missing data. All outcome variables, except for ventilation in the most dependent region of the lungs in humans and V_T during continuous airway pressure in pigs, were log-transformed to obtain a normal distribution of residuals. P values were corrected for multiple comparisons using the Bonferroni method; a two-sided corrected P value less than 0.05 was considered statistically significant. In humans, we performed 21 comparisons (pairwise combination of seven levels). Therefore, the target P value was 0.002 (i.e., 0.05/21). We presented instead the “corrected P value,” i.e., the unadjusted P value × 21. The same method was applied for pigs, where we performed 10 comparisons (pairwise combination of five levels). For the analysis of heart rate and blood pressure, which were collected in two time points, we performed a Wilcoxon signed-rank test.

All figures were plotted, and statistical analyses were performed in software R 3.5.1 (R Foundation for Statistical Computing, Austria) and Rstudio (Rstudio Team, USA).

Six pigs and nine patients were successfully included in this study. All values are presented as median [interquartile range‚ 25th to 75th percentile]. Please refer to Supplemental Digital Content 1 (https://links.lww.com/ALN/C800) and Supplemental Digital Content 6 (https://links.lww.com/ALN/C832) for further information.

We successfully blocked the phrenic nerve bilaterally in six pigs (100%). All variables but left lung ventilation decreased significantly when compared from baseline to 10 min after left phrenic block (figs. 3 and 4; table 1): electrical activity of the diaphragm from 20.5 µV [19.8 to 27.5] to 0.7 µV [0.6 to 2.2] (corrected P < 0.001); ΔP_eso from 10.5 cm H₂O [8.9 to 10.9] to 2.7 cm H₂O [2.4 to 3.2] (corrected P < 0.001); V_T from 7.4 ml/kg [7.2 to 8.4] to 5.9 ml/kg [5.5 to 6.7] of predicted body weight (corrected P < 0.001); peak transpulmonary pressure from 25.8 cm H₂O [20.2 to 27.2] to 17.7 cm H₂O [13.8 to 18.8] (corrected P < 0.001); and driving pressure from 28.7 cm H₂O [20.4 to 30.8] to 19.4 cm H₂O [15.2 to 22.9] (corrected P < 0.001). The percentage of ventilation in the most dependent region of the lung decreased from 29.3% [26.4 to 29.5] to 20.1% [15.3 to 20.8] (corrected P < 0.001), and the percentage of ventilation distribution to the left lung ventilation did not change from 38.5 [37.8 to 41.4] to 37.0 [33.7 to 46.2] (corrected P > 0.999).

In four pigs ventilated on continuous positive airway pressure (fig. 5), there was a decrease in V_T from 5.9 ml/kg [4.9 to 7.2] to 3.1 ml/kg [2.0 to 4.0] of predicted body weight (corrected P = 0.009) when baseline was compared to 10 min after right phrenic block. At baseline, four pigs had pendelluft. After bilateral phrenic nerve block, pendelluft was reduced from 8% [6.0 to 8.6] to 0% [0 to 0.0] (P < 0.001) of V_T before the ventilator was triggered (fig. 4C).

Nine patients (three women and six men) with ARDS were included in this study. Table 2 shows patient characteristics and respiratory variables after pressure support titration at baseline.

The phrenic nerves could be visualized bilaterally in all patients. Bilateral phrenic nerve block was then successfully performed in all patients with the administration of 15 ml lidocaine, 2%, around each phrenic nerve. The left phrenic nerve block required 6.5 min [4.5 to 12.2] (between baseline and 4 min after left phrenic block), and the right phrenic block required 7.4 min [5.8 to 11.1] (between the 10 min after left phrenic block and 4 min after right phrenic block time points). From baseline to bilateral phrenic nerve block, the time was 14.6 [11.3 to 23.4] min.

No complications such as severe tachycardia or prolonged phrenic nerve palsy were observed in any participant. In addition, none of our patients presented elevation of diaphragm the day after the phrenic nerve block.

After bilateral phrenic nerve block and compared to baseline (figs. 6 and 7; table 3), there was a decrease (10 min after right phrenic block) in V_T from 8.2 ml/kg [7.9 to 11.1] to 6.0 ml/kg [5.7 to 6.7] of predicted body weight (corrected P < 0.001) and peak transpulmonary pressure from 24.1 cm H₂O [20.1 to 30.4] to 18.3 cm H₂O [14.4 to 26.5] (corrected P = 0.025).

One hour after right phrenic block, when compared to baseline (figs. 6 and 7; table 3), V_T decreased from 8.2 [7.9 to 11.1] to 6.5 [6.1 to .9] (corrected P < 0.001), but peak transpulmonary pressure did not decrease (24.1 cm H₂O [20.1 to 30.4] to 21.7 cm H₂O [16.3 to 29.3] corrected P > 0.999). In the following day (i.e., the final time point), V_T of 7.7 ml/kg [7.2 to 9.6] of predicted body weight and peak transpulmonary pressure of 31.9 cm H₂O [22.1 to 43.7] were not different from baseline (corrected P > 0.999 for both).

Bilateral phrenic nerve block (10 min after right phrenic block) compared to baseline (figs. 6 and 7) decreased electrical activity of the diaphragm from 20 [12.2 to 29.9] to 0.3 [0.2 to 1.0] µV (corrected P < 0.001), driving pressure from 24.7 cm H₂O [20.4 to 34.5] to 18.4 cm H₂O [16.8 to 20.7] (corrected P < 0.001), ΔP_eso from 7.8 cm H₂O [5.6 to 9.8] to 1.9 cm H₂O [1.6 to 3.3] (corrected P < 0.001), and percentage of ventilation in the most dependent region from 11.5% [8.5 to 12.6] to 7.9% [5.3 to 8.6] (corrected P = 0.005). There was no difference in respiratory rate (27 breaths/min [24 to 30] to 25 breaths/min [22 to 30]; corrected P > 0.999) and percentage of left lung ventilation (45.7% [37.3 to 47.1] to 47.4% [39.0 to 47.8]; corrected P > 0.999).

One hour after bilateral phrenic nerve block, electrical activity of the diaphragm (20 µV [12.2 to 29.9] to 8.4 µV [1.5 to 17.4]; corrected P = 0.062) and ΔP_eso (7.8 cm H₂O [5.6 to 9.8] to 4.9 cm H₂O [3.9 to 5.6]; corrected P = 0.230) were not different from baseline, while driving pressure decreased from 24.7 cm H₂O [20.4 to 34.5] to 19.4 cm H₂O [18.4 to 24.8] (corrected P < 0.001). The respiratory rate did not change (27 breaths/min [24 to 30] to 24 breaths/min [23 to 25]; corrected P > 0.999), as well as the percentage of ventilation in the most dependent region of the lung (11.5% [8.5 to 12.6] to 7.5% [3.6 to 13.4]; corrected P > 0.999) and left lung ventilation (45.7% [37.3 to 47.1] to 45.7% [37.8 to 47.0]; corrected P > 0.999). On the following day (i.e., the final time point), electrical activity of the diaphragm of 14.8 µV [11.9 to 23.3], driving pressure of 23.8 cm H₂O [20.1 to 33.4], ΔP_eso of 6.5 cm H₂O [5.9 to 20.5], respiratory rate of 25 breaths/min [20 to 28], percentage of ventilation in the most dependent region of the lung of 10.2% [5.2 to 13.72], and left lung ventilation of 39.3% [36.9 to 43.6] were not different from baseline (corrected P > 0.999 for all variables).

There was no change from baseline to bilateral phrenic nerve block in heart rate (89 beats/min [79 to 92] vs. 95 beats/min [89 to 101]; [95% CI, –3.1 to 16.6]; P = 0.164; fig. 8A) and mean blood pressure (87 mmHg [83 to 90] vs. 99 mmHg [87 to 103]; [95% CI, −4.6 to 19.0]; P = 0.213; fig. 8B).

For one patient, we did not acquire data during the following day. For another patient, deep sedation and neuromuscular blocking agents were reinstated due to strong inspiratory effort before full recovery. A third patient had sedation raised due to agitation during the night, and thus, ΔP_eso did not recover. Finally, another patient presented activation of expiratory abdominal muscle in the final time point, a condition that was not present at the beginning of the protocol; therefore, his peak transpulmonary pressure and ΔP_eso at the final time point were higher than at baseline. All these patients were included in the final analysis. Of note, all these patients were eventually weaned from the ventilator. Finally, none of our patients presented elevation of diaphragm the day after the phrenic nerve block.

Electrical activity of the diaphragm, peak transpulmonary pressure, and ΔP_eso fully recovered from bilateral phrenic nerve block within 1 h. V_T and driving pressure recovered from bilateral phrenic nerve block in 12.7 h [6.7 to 13.7].

In this proof-of-concept study, we evaluated a novel strategy to deliver lung-protective ventilation in six pigs and in nine ARDS patients without the use of deep sedation and systemic neuromuscular blocking agents. We found that (1) bilateral phrenic nerve block was feasible, safe, and effective in the short term; (2) it reduced electrical activity of the diaphragm, V_T, ΔP_eso, peak transpulmonary pressure, and driving pressure without modifying respiratory rate; and (3) all parameters fully recovered after an average of 13 h.

Excessively strong respiratory efforts have the potential to be injurious for the lung and the diaphragm.²⁸  There is no consensus, however, on managing patients with vigorous spontaneous inspiratory efforts. Animal studies have demonstrated the existence of self-inflicted lung injury,^9,10,29  but no human data yet have shown if and when it occurs.^6,29–31  Case reports or small series have associated strong inspiratory efforts with worse clinical outcomes in patients but do not demonstrate causality.^8,32,33  Before phrenic nerve blockade, pigs and patients had V_T, driving pressure, or both above the limits of lung protection recently proposed.²⁸  Moreover, pigs had electrical activity of the diaphragm and ΔP_eso comparable to other studies that associated vigorous inspiratory efforts and self-inflicted lung injury.^6,29 

After bilateral block, there was a decrease in V_T toward values within purported lung protective thresholds. Moreover, driving pressure and peak transpulmonary pressure, an indication of maximal lung stress, are dependent from the set level of inspiratory pressure support and significantly decreased from baseline to bilateral block. The decrease of peak transpulmonary pressure in our study is lower than that described in patients with partial neuromuscular blockade by Doorduin et al.¹²  Airway driving pressure is associated to mortality in ARDS³⁴  and assisted ventilation.³⁵  Recent studies showed that driving pressure can be reliably measured during pressure support ventilation.^23,36  Airway driving pressure is a surrogate for lung cyclic strain,³⁷  defined as the ratio between tidal volume and functional residual capacity. Available data suggest that the prognostic values of airway and transpulmonary driving pressures seem similar.³⁸  Finally, airway driving pressure is easier to measure and offers less technical difficulties than transpulmonary driving pressure. Hence‚ a strategy that lowers airway driving pressure or tidal volume may benefit patients.

One hour after bilateral phrenic nerve block, electrical activity of the diaphragm, ΔP_eso, and peak transpulmonary pressure had fully recovered, while V_T and driving pressure had recovered only partially. It is possible that the decrease in minute ventilation increased Paco₂ and stimulated the respiratory drive, which ultimately increased the use of the diaphragm and other respiratory muscles, offsetting the phrenic nerve block. Unfortunately, we did not use a specific clinical scale to measure respiratory distress, or assess accessory muscle activity. In addition to likely Paco₂ accumulation, the effect of lidocaine may have partly weaned off 1 h after bilateral block.³⁹  Taken together, both conditions may have contributed to an early partial recovery.

Ventilation in the most dependent part of the lung decreased after phrenic nerve paralysis in pigs and humans. Phrenic nerve block suppressed diaphragmatic contraction, thus reducing ventilation near the diaphragm area. Bilateral phrenic nerve block could favor lung collapse, a consequence that may be prevented with PEEP. Moreover, the suppression of the diaphragmatic contraction abolished pendelluft (i.e., an internal redistribution of tidal ventilation), a phenomenon related to regional distension and adverse outcomes.^22,40  Finally, sequential phrenic nerve block did not cause left lung to right lung pendelluft.

Unlike partial neuromuscular blockade,¹²  our technique does not need sedatives, analgesics, or neuromuscular blocking agents. The two patients monitored with Bispectral Index presented levels between 70 and 85, compatible with light sedation. Half of our patients were under mild to moderate sedation; one was off continuous sedation and had a Richmond Agitation–Sedation Scale score of −1 (table 1).

During resting breathing in healthy subjects, the diaphragm is responsible for 60 to 80% of the inspiratory pressure, while the accessory muscles are responsible for the remaining activity.⁴¹  Lidocaine blocks the propagation of the action potential, hence reducing electrical activity of the diaphragm, but does not interfere with other respiratory muscles due to their different innervation, which explains the residual ΔP_eso seen in pigs and in patients. The remaining ΔP_eso was able to trigger the ventilator, allowing patients and pigs to continue under assisted ventilation.

Patients on invasive mechanical ventilation are at high risk for diaphragm myotrauma, a condition that may be a result of overassistance, load-induced injury, eccentric contractile injury, and excessive shortening.⁴²  Such associations were evidenced through mediation analysis and need further confirmation in prospective studies. Despite controversial results, a common strategy to temporarily reduce diaphragm myotrauma is to use paralyzing agents.^2,43  However, systemic paralyzing agents may contribute to intensive care unit–acquired weakness and general muscle atrophy. Our novel method could maintain patients under protective mechanical ventilation while avoiding deep sedation and paralyzing agents. To abolish electrical activity of the diaphragm and therefore diaphragm contraction with phrenic nerve block, however, may hold the same risks as using systemic paralyzing agents referring to diaphragm atrophy. This risk of diaphragm atrophy may be diminished if it is possible to titrate a continuous infusion of local anesthetics through a perineural catheter to reduce, but not abolish, electrical activity of the diaphragm. Moreover, phrenic nerve block paralyzes the diaphragm, but does not paralyze other muscles, which may prevent accessory muscle atrophy.⁴⁴ 

Our approach takes less than 15 min to perform with a trained anesthesiologist. It reduces peak transpulmonary pressure, V_T, and driving pressure and can be performed in patients under light to moderate sedation. Altogether, it may be possible to keep a patient under protective assisted ventilation while allowing for participation in physical and occupational therapy, an intervention that is associated with better functional outcomes.⁴⁵  Furthermore, since the intervention is a superficial peripheral ultrasound-guided nerve block done in an area susceptible to compression, anticoagulation is not a contraindication.⁴⁶ 

First, only female pigs were included in this protocol, which could limit the results. However, there are no studies associating sex of animals with clinically significant differences in respiratory physiology and anatomy. Second, the current study was a single-center, proof-of-concept, feasibility study. We did not aim to detect patient-centered outcomes, but rather surrogate physiologic variables associated with ventilator-induced lung injury. Third, the duration of the block was longer than expected. There is a wide variation in the length of lidocaine-induced motor paralysis,^47,48  but no studies have administered the same dose of lidocaine we used in the cervical portion of the phrenic nerve, where there is less room for lidocaine dispersion than in a supraclavicular block, for instance. Finally, the lidocaine dose administered may have been higher than necessary, further extending recovery periods. Fourth, the use of ultrasound imaging and a neurostimulator for phrenic nerve block requires training, which may limit this approach. Fifth, we did not routinely measure Paco₂ in pigs or in patients and did not assess the use of extra-diaphragmatic inspiratory muscles or activation of expiratory muscles. Therefore, the effects of bilateral phrenic nerve block on Paco₂ and muscle activity need further investigation. Finally, although we cannot comment on the recovery of three of our patients, they were all weaned off the ventilator.

In conclusion, the current proof-of-concept study demonstrates that bilateral phrenic nerve block can be feasible and efficient in the short term, and may reduce strong spontaneous inspiratory efforts. This technique might reduce the need for deep sedation and neuromuscular blocking agents in patients under mechanical ventilation.

The authors thank Takeshi Yoshida, M.D., Ph.D. (Department of Anesthesiology and Intensive Care Medicine, Osaka University Graduate School of Medicine, Osaka, Japan), and Susimeire Gomes, Ph.D. (Division of Pulmonology [Laboratory of Medical Investigation 09], Faculty of Medicine, University of São Paulo, São Paulo, Brazil) for their insights in designing the study, as well as data acquisition and analysis.

Dr. Costa has received financial support from Timpel S.A. (São Paulo, Brazil). Dr. Brochard has received financial support from Medtronic (Minneapolis, Minnesota), Draeger (Lubeck, Germany), and Fisher Paykel (Auckland, New Zeland), as well as equipment from Sentec (Therwill, Switzerland).

The authors declare no competing interests.