Vigorous spontaneous effort can potentially worsen lung injury. This study hypothesized that the prone position would diminish a maldistribution of lung stress and inflation after diaphragmatic contraction and reduce spontaneous effort, resulting in less lung injury.
A severe acute respiratory distress syndrome model was established by depleting surfactant and injurious mechanical ventilation in 6 male pigs (“mechanism” protocol) and 12 male rabbits (“lung injury” protocol). In the mechanism protocol, regional inspiratory negative pleural pressure swing (intrabronchial balloon manometry) and the corresponding lung inflation (electrical impedance tomography) were measured with a combination of position (supine or prone) and positive end-expiratory pressure (high or low) matching the intensity of spontaneous effort. In the lung injury protocol, the intensities of spontaneous effort (esophageal manometry) and regional lung injury were compared in the supine position versus prone position.
The mechanism protocol (pigs) found that in the prone position, there was no ventral-to-dorsal gradient in negative pleural pressure swing after diaphragmatic contraction, irrespective of the positive end-expiratory pressure level (–10.3 ± 3.3 cm H2O vs. –11.7 ± 2.4 cm H2O at low positive end-expiratory pressure, P = 0.115; –10.4 ± 3.4 cm H2O vs. –10.8 ± 2.3 cm H2O at high positive end-expiratory pressure, P = 0.715), achieving homogeneous inflation. In the supine position, however, spontaneous effort during low positive end-expiratory pressure had the largest ventral-to-dorsal gradient in negative pleural pressure swing (–9.8 ± 2.9 cm H2O vs. –18.1 ± 4.0 cm H2O, P < 0.001), causing dorsal overdistension. Higher positive end-expiratory pressure in the supine position reduced a ventral-to-dorsal gradient in negative pleural pressure swing, but it remained (–9.9 ± 2.8 cm H2O vs. –13.3 ± 2.3 cm H2O, P < 0.001). The lung injury protocol (rabbits) found that in the prone position, spontaneous effort was milder and lung injury was less without regional difference (lung myeloperoxidase activity in ventral vs. dorsal lung, 74.0 ± 30.9 μm · min–1 · mg–1 protein vs. 61.0 ± 23.0 μm · min–1 · mg–1 protein, P = 0.951). In the supine position, stronger spontaneous effort increased dorsal lung injury (lung myeloperoxidase activity in ventral vs. dorsal lung, 67.5 ± 38.1 μm · min–1 · mg–1 protein vs. 167.7 ± 65.5 μm · min–1 · mg–1 protein, P = 0.003).
Prone position, independent of positive end-expiratory pressure levels, diminishes a maldistribution of lung stress and inflation imposed by spontaneous effort and mitigates spontaneous effort, resulting in less effort-dependent lung injury.
Prone positioning during mechanical ventilation for patients with acute lung injury has been shown to increase oxygenation and possibly improve outcome
It is now widely used for patients with COVID-19 failing routine ventilation protocols
Its use during spontaneous ventilation has increased as result of the pandemic, yet detailed data on its ventilatory effects have not been well established
The authors utilized porcine and rabbit models of lung injury to evaluate pulmonary mechanics, distribution of ventilation, and biochemical and histologic effects on lung injury with varying positive end-expiratory pressure levels
Independent of positive end-expiratory pressure levels, prone positioning reduced maldistribution of lung stress and reduced effort-dependent evidence of lung injury
Spontaneous breathing using respiratory muscles is physiologically normal and therefore has been traditionally facilitated during mechanical ventilation.1 Negative deflection (“swing”) in pleural pressure resulting from diaphragmatic contraction is evenly transmitted across the whole lung surface, creating a uniform increase in transpulmonary pressure at any given airway pressure (Paw): this is called “fluid-like” behavior.2 Thus, spontaneous breathing achieves homogeneous inflation at lower levels of Paw during mechanical ventilation, improving ventilation/perfusion and gas exchange, and preserving diaphragm function.1,3 Although such benefits of spontaneous breathing have been reported during mechanical ventilation, it may also potentially injure the lungs and diaphragm when spontaneous effort is vigorous and/or when lung injury is severe.3–6
In the severely injured lung, negative deflection in pleural pressure resulting from diaphragmatic contraction is partially used on local lung deformation (i.e., dense, atelectatic area resisting dynamic shape changes) and thus is not evenly transmitted to the entire lung; this is called “solid-like” behavior.2,7 Such a maldistribution of lung stress imposed by spontaneous breathing is known to cause injurious inflation patterns (e.g., pendelluft, local volutrauma2,7 ). In addition, several factors increase the strength and injury potential of spontaneous breathing effort in severe acute respiratory distress syndrome (ARDS), including acidemia, hypercapnia, and hypoxemia,8 as well as reduced lung volume due to dorsal atelectasis.9,10
Turning to the prone position gravitationally translocates atelectasis (dense solid-like lung tissue resisting dynamic shape changes) from the dorsal to ventral lung, as is obvious from previous studies.11 Because the dorsal lung (facing muscular parts of diaphragm) is now open and less solid-like atelectatic in the prone position, it might diffuse the inspiratory stress after diaphragmatic contraction from being local and injurious to generalized and less injurious (e.g., less pendelluft, less local volutrauma). In several studies, the prone position is also shown to have the similar effect of recruiting lung and increasing lung volume as higher positive end-expiratory pressure (PEEP).11,12 Lung recruitment may minimize the injurious effect of spontaneous effort (e.g., large tidal volume, high transpulmonary pressure) by increasing lung volume, shortening diaphragm length, and thereby generating less force from the diaphragm.6,9,10,13–15
The prone position has been traditionally used under passive conditions (e.g., more than 90% of patients in the prone position received muscle paralysis for more than 5 days),16 and the interaction of the prone position with spontaneous breathing has not been evaluated well in severe ARDS. Based on this reasoning, we hypothesized that if spontaneous effort is permitted while in the prone position, it would diminish a maldistribution of lung stress and inflation imposed by spontaneous effort and decrease spontaneous effort, resulting in less lung injury.
We tested this hypothesis in established models of severe ARDS. First, in the “mechanism” protocol using pigs, to evaluate regional lung stress and the corresponding inflation pattern caused by spontaneous effort, we measured the impact of PEEP (high and low) and position (supine and prone) on regional lung inflation (electrical impedance tomography in pigs) and regional inspiratory negative pleural pressure swings (intrabronchial balloon manometry2,17 in pigs). Second, in the “lung injury” protocol using rabbits, we measured the impact of position on the strength of spontaneous effort (negative deflection in esophageal pressure [Peso] in rabbits), and on regional injury associated with spontaneous effort (total protein in bronchoalveolar lavage, lung myeloperoxidase activity in rabbits).
Materials and Methods
Two series of animal experiments (pigs, rabbits) were conducted from 2017 through 2018 (before the COVID-19 pandemic), both approved by the Animal Care Committee of the Hospital for Sick Children in Toronto (Toronto, Ontario, Canada; approval No. 45697). The animals were cared for in accordance with the hospital’s standards for the care and use of laboratory animals.
Series 1 Mechanism Protocol: Anesthetized Pig Experiments
The schematic of study protocol is described in figure 1A. Six male Yorkshire pigs (n = 6; 30.9 to 39.3 kg) were anesthetized with 7 mg · kg–1 · h–1 ketamine and 2 mg · kg–1 · h–1 propofol and tracheostomized. Negative toe pinch was confirmed throughout the protocol. An esophageal balloon catheter (NutriVent, Sidam, Italy) was inserted to measure Peso, filled with 1.0 ml as a minimal nonstress volume, and calibrated.18 Neuromuscular blockade rocuronium bromide boluses of 0.5 mg · kg–1 were used to prevent spontaneous breathing effort when necessary.
Experimental lung injury was induced in the supine position by repeated saline lung lavage (30 ml · kg–1, 37°C),19 and surfactant depletion was considered stable when the Pao2/fractional inspired oxygen tension (Fio2) ratio was less than 100 mmHg for 10 min, at a PEEP of 5 cm H2O. Injurious mechanical ventilation was commenced and continued for 60 min using assisted pressure control: Fio2, 1.0; rate, 25 breaths/min; and pressure trigger, –2 cm H2O (Servo 300, Siemens-Elema AB, Sweden). Ventilator-induced lung injury was induced with the following driving pressure/PEEP combinations adjusted every 15 min to maintain Pao2 of greater than 55 to 65 mmHg: 41/1, 39/3, 37/5, 35/7, 33/9, 31/11, or 29/13 cm H2O.13
The animals were then randomly assigned to four acquisition periods (each period comprised high or low PEEP and supine or prone position):
Low PEEP, supine
High PEEP, supine
Low PEEP, prone
High PEEP, prone
Randomization was from a bag of coded letters. Static respiratory system compliance was measured with decremental PEEP steps (modified from Yoshida et al.20 ), starting at a PEEP of 20 cm H2O and reducing by 2 cm H2O every 30 s until an oxygen saturation measured by pulse oximetry of approximately 90% was reached. Ventilation was set at Fio2 1.0, inspiratory pressure was set at 15 cm H2O, and the respiratory rate was 40 breaths/min. At a PEEP of 20 cm H2O, the Pao2/Fio2 ratio was approximately 400 mmHg in all animals. High and low PEEP were defined as follows:
High PEEP is the PEEP at which respiratory system compliance is maximal after decremental PEEP steps
Low PEEP is the PEEP at which oxygen saturation measured by pulse oximetry is approximately 90% (Pao2 is approximately 60 mmHg)
The lungs were fully recruited in the supine position to homogenize lung volume history before randomization to each acquisition period and ventilated for approximately 15 min for stabilization. In each acquisition period, low tidal volume (VT) ventilation employed assisted volume-controlled ventilation: VT, 7 ml · kg–1; rate, 30 breaths/min; inspiratory to expiratory ratio, 1:2 (no inspiratory pause); pressure trigger, –2 cm H2O; and Fio2, 1.0.
At the start of each acquisition period, the absence of respiratory effort was confirmed by a lack of negative deflection in Peso. Spontaneous breathing effort was subsequently facilitated by adding carbon dioxide (up to 0.10) until a negative swing in Peso of –10 cm H2O was reached. It usually took approximately 30 min to reach the target value of Peso. The animals were sacrificed with IV sodium pentobarbital.
Electric Impedance Tomography
In all animals (n = 6), electrical impedance tomography data were recorded (PulmoVista 500, Dräger, Germany) continuously during the spontaneous breathing titration period (from paralysis to Peso of –10 cm H2O). Local lung inflation was analyzed after division of the image into four equal zones, from zone 1 (most ventral) to zone 4 (most dorsal), where each zone comprised 25% of the ventrodorsal distance and encompassed the complete area of the lung encircled by the band. We considered zone 1 (the most ventral one) and zone 4 (the most dorsal one) as representative of ventral lung and dorsal lung to be analyzed, respectively. The magnitude of local lung inflation imposed by spontaneous effort was estimated by the size of passive VT during muscle paralysis to achieve the same degree of local lung inflation.2 This estimation in each sequence was performed when ΔPeso was –10 cm H2O (i.e., the same intensity of spontaneous effort). After measuring the magnitude of local lung inflation (represented by ΔZ in electrical impedance tomography) when ΔPeso was –10 cm H2O at a fixed, global VT of 7 ml/kg during assisted volume-controlled ventilation, we paralyzed the animal and started to increase VT setting during volume-controlled ventilation, until the same magnitude of local lung inflation (represented by ΔZ in electrical impedance tomography) developed in the dorsal lung.
Pleural Pressure Measurement
The local negative swing in pleural pressure was determined in nondependent and dependent regions (one pig did not survive; n = 5) by balloon catheter occlusion of subsegmental bronchi via a fibreoptic bronchoscope, as follows: nondependent region, left B; dependent region: left lower lobe beyond D4. The occluded subsegments were connected to a differential pressure transducer through the intrabronchial balloon catheter without airflow influx, thereby allowing continuous measurement of changes in occluded subsegment pressure. The pressure swings in the occluded subsegments were used as surrogates for negative pleural pressure swings, as described previously.2,17 The occluded lung regions were filled with air until the alveolar pressure inside each target subsegmental region reached 20 (or 30) cm H2O in nondependent (and dependent) lung regions, respectively, assuming that this opening pressure was sufficient to recruit the occluded lung regions. Simultaneous pressure recording of negative pleural pressure swings and ΔPeso were performed, while preserving spontaneous effort. All measurements were performed when ΔPeso was –10 cm H2O.
Series 2 Lung Injury Protocol: Anesthetized Rabbit Experiments
A schematic of study protocol is shown in figure 1B. Twelve New Zealand white rabbits (adult, male, 2.9 to 3.9 kg) were anesthetized with intravenous propofol (10 to 100 mg · kg–1 · h–1) and ketamine (1 to 5 mg · kg–1 · h–1) and tracheostomized. Negative toe pinch was confirmed throughout the protocol. An esophageal balloon (SmartCath, Bicore, USA) was inserted to measure Peso and filled with air (0.3 ml as minimal nonstress volume), and its position was verified.18
Experimental lung injury was induced in the supine position by repeated lung lavage,19 and surfactant depletion was considered stable when the Pao2/Fio2 ratio was less than 150 mmHg for 10 min at a PEEP of 3 cm H2O. Injurious mechanical ventilation using assisted pressure control consisted of VT of approximately 15 ml · kg–1 (by adjusting inspiratory pressure), and a PEEP of 2 cm H2O. PEEP was adjusted (increased or decreased) by 2 cm H2O to maintain a Pao2/Fio2 of 55 to 65 mmHg after 15 min and continued for 30 min.
The lungs were fully recruited in the supine position, and PEEP was set at where the Pao2/Fio2 ratio was approximately 100 mmHg in the supine position. Then the animals were randomly assigned to one of two groups (n = 6 for each group):
Supine plus spontaneous breathing
Prone plus spontaneous breathing
Randomization was from a bag of coded letters. The animals were then ventilated for 4 h using low VT ventilation, using pressure-controlled ventilation: VT, 6 ml · kg–1 (by adjusting inspiratory pressure); respiratory rate, 60 to 120 breaths/min (targeted to Paco2 of less than 50 mmHg); inspiratory time, 0.2 s; minimum flow trigger; and Fio2 adjusted to target Pao2 of 100 mmHg. All of the animals (n = 12) survived the protocol. After 4 h of mechanical ventilation, the animals were sacrificed with IV sodium pentobarbital, and the lungs were excised.
Wet to Dry Lung Weight
The right upper and middle lobes of the lung were weighed, placed in a warming oven (37°C), and weighed daily until the weight was stable.
Bronchoalveolar fluid was collected from the left whole lung by injecting 10 ml of normal saline three times; then the total protein in the bronchoalveolar fluid was quantified. Lung myeloperoxidase activity was measured21 from lung biopsies; a lung tissue sample (8 × 8 × 8 mm) was taken from the nondependent and dependent right middle lobes. One investigator (G.O.), who was blind to sampling regions and group allocation, performed the analysis.
The right lower lobe was fixed with intratracheal insufflation of 10% formalin of 15 ml for at least 24 h. The right lower lobe was sectioned transversely (5-mm slices) and embedded in paraffin. In addition, 3-μm slices were stained with hematoxylin and eosin. Representative histologic images in each group are presented.
The definitions of pulmonary pressures are as follows:
Negative swing in Peso: ΔPeso was determined from the amount of decrease (spontaneous breathing) in Peso from the start of inspiration.
Negative swing in pleural pressure: Δ pleural pressure was determined from the amount of decrease (spontaneous breathing) in pleural pressure from the start of inspiration.
Maximal (inspiratory) transpulmonary pressure: Peak transpulmonary pressure equaled the maximal value of [Paw – Peso] cm H2O, usually corresponding to the time of the most negative value of Peso (maximum inspiration).
Plateau (inspiratory) transpulmonary pressure: Plateau transpulmonary pressure equaled [plateau Paw – end-inspiratory Peso] cm H2O.
Plateau pressure: Paw measured during a short inspiratory hold (i.e., zero flow phase).
Driving pressure equaled [plateau Paw – PEEP] cm H2O.
Peak Δ transpulmonary pressure: Peak Δ transpulmonary pressure equaled [Paw – PEEP – (ΔPeso)] cm H2O, corresponding to the time of maximal value of peak Δ transpulmonary pressure.
Compliance of the respiratory system equaled [VT/(driving pressure)] mL · cm H2O–1.
Statistical analyses were performed using SPSS13.0 for Windows (SPSS, USA). The study was exploratory, and the sample size was not formally calculated, but it was based on experience. Normal distribution of data was checked with histography. The results are expressed as mean ± SD. One-way ANOVA was used to compare myeloperoxidase activities among regions. Two-way ANOVA with repeated measures evaluated the effects of time and group on respiratory variables. Two-way ANOVA was applied to evaluate the effects of lung regions (ventral vs. dorsal) and condition differences on lung stress and lung inflation imposed by spontaneous effort. In the post hoc analysis, a Dunnett’s test was used to compare repeated values with the value at the start of the protocol (i.e., 0 h), and Tukey’s pairwise multiple comparison test was used to determine condition differences. Unpaired t tests were used to compare the wet to dry ratio and bronchoalveolar fluid protein. All tests were two-tailed, and differences were considered significant when P < 0.05.
Mechanism Protocol in the Anesthetized Pig
VT was low and similar (volume-controlled ventilation: 6.7 ± 0.6 to 6.9 ± 0.5 ml/kg) in all four conditions (“condition” P = 0.772 by two-way repeated ANOVA) at baseline (paralyzed) and throughout titration of spontaneous effort (“time” P = 0.081 by two-way repeated ANOVA; Supplemental Digital Content table S1, http://links.lww.com/ALN/C801). The development of spontaneous breathing did not alter global VT (as anticipated, given the volume-controlled ventilation). The swing (deflection) in esophageal pressure (ΔPeso) increased until it reached –10 cm H2O during spontaneous effort titration as per protocol in all groups (Supplemental Digital Content table S1, http://links.lww.com/ALN/C801).
Local Pleural Pressure during Spontaneous Effort
The regional distribution of pleural pressure (fig. 2) was measured and evaluated under the same amount of spontaneous effort in all conditions (i.e., ΔPeso = –10 cm H2O). The magnitude of negative inspiratory pleural pressure in the dorsal (dependent) lung was almost twofold greater than negative inspiratory pleural pressure in the ventral (nondependent) lung at low PEEP in the supine position (Δ pleural pressure in ventral vs. dorsal lung: –9.8 ± 2.9 cm H2O vs. –18.1 ± 4.0 cm H2O; P < 0.001; fig. 2A). High PEEP in the supine position significantly reduced a ventral to dorsal gradient in inspiratory Δ pleural pressure (dorsal Δ pleural pressure in low PEEP vs. high PEEP: –18.1 ± 4.0 cm H2O vs. –13.3 ± 2.3 cm H2O; P < 0.001; fig. 2A,vs. fig. 2B). In the prone position, however, there was no ventral to dorsal gradient in local Δ pleural pressure after diaphragmatic contraction, irrespective of the PEEP level (Δ pleural pressure in ventral [dependent] vs. dorsal [nondependent] lung: –10.3 ± 3.3 cm H2O vs. –11.7 ± 2.4 cm H2O at low PEEP, P = 0.115; –10.4 ± 3.4 cm H2O vs. –10.8 ± 2.3 cm H2O at high PEEP, P = 0.715; fig. 2, C and D).
Local Lung Inflation during Spontaneous Effort versus Muscle Paralysis
When comparing the regional distribution of lung inflation, the strength of spontaneous effort was matched among animals (i.e., ΔPeso equaled –10 cm H2O) under a fixed global VT (volume-controlled mode: approximately 7 ml/kg, Supplemental Digital Content table S1, http://links.lww.com/ALN/C801). Local distribution of lung inflation imposed by spontaneous effort in electrical impedance tomography reflected the ventral to dorsal gradient in negative Δ pleural pressure during diaphragmatic contraction.
In the supine position during low PEEP, spontaneous effort increased local lung inflation in the dependent (dorsal) lung, in the same region where more negative Δ pleural pressure was localized (fig. 2A). At low PEEP and supine, a significantly larger passive VT (15.4 ± 2.3 ml/kg, P = 0.009 vs. high PEEP and supine, P < 0.001 vs. high PEEP and prone, P < 0.001 vs. low PEEP and prone) was required to achieve inspiratory inflation of the dependent lung (fig. 2A) comparable to that achieved during spontaneous effort, despite limiting global VT to approximately 7 ml/kg. The magnitude of local dependent lung inflation imposed by spontaneous effort was significantly less at high PEEP and supine (P = 0.009 vs. low PEEP and supine), and thus the distribution of lung inflation was similar among lung regions (passive VT required in ventral [nondependent] vs. dorsal [dependent] lung: 6.2 ± 4.9 ml/kg vs. 11.2 ± 2.6 ml/kg; P = 0.062; fig. 2B). In the prone position, the distribution of lung inflation was not altered by spontaneous effort at low PEEP (passive VT required in ventral [dependent] vs. dorsal [nondependent] lung: 7.3 ± 2.6 ml/kg vs. 6.1 ± 2.2 ml/kg; P = 0.512; fig. 2C) and at high PEEP (passive VT required in ventral [dependent] vs. dorsal [nondependent] lung: 7.1 ± 2.2 ml/kg vs. 7.0 ± 2.0 ml/kg; P = 0.943; fig. 2D).
Lung Injury Protocol in the Anesthetized Rabbit
The dose of propofol and ketamine was similar in the supine position versus the prone position (propofol: 19 ± 5 mg · kg–1 · h–1vs. 22 ± 5 mg · kg–1 · h–1, P = 0.394; ketamine: 3.1 ± 1.9 mg · kg–1 · h–1vs. 2.1 ± 0.7 mg · kg–1 · h–1, P = 0.273). The values of VT (approximately 6 ml/kg) were similar in both groups (“group” P = 0.853 by two-way repeated ANOVA) throughout the protocol (“time” P = 0.837 by two-way repeated ANOVA). Oxygenation (Pao2/Fio2) was greater during spontaneous effort in the prone versus supine position (group P = 0.008 by two-way repeated ANOVA; table 1). In the supine position, oxygenation increased transiently for approximately the first hour after commencement of spontaneous breathing and decreased thereafter (table 1). Respiratory system compliance decreased over time in the supine position with spontaneous effort but did not decrease in the prone position with spontaneous effort. Respiratory system compliance was higher after 2 h in the prone position vs. supine position (respiratory system compliance at 4 h: 2.1 ± 0.9 ml/cm H2O vs. 1.1 ± 0.2 ml/cm H2O; P = 0.034; table 1).
Spontaneous Effort in Supine versus Prone Position
The intensity of spontaneous effort in terms of frequency (estimated by respiratory rate) and magnitude (estimated by negative ΔPeso) was significantly less in the prone versus supine groups (table 1; fig. 3A) despite the use of the same doses of sedatives, the maintenance of constant Pao2 (approximately 100 mg by adjusting Fio2), and the same value of Paco2 (table 1). The deflections in ΔPeso became significantly more negative in the supine position but remained constant in the prone position (ΔPeso at 4 h: –3.9 ± 1.3 cm H2O vs. –1.6 ± 1.1 cm H2O; P = 0.008; fig. 3A). Spontaneous respiratory rate (and thus minute ventilation) was significantly higher in the supine vs. prone groups (table 1). At all times during spontaneous breathing after time zero, the peak Δ transpulmonary pressure (at maximum inspiration) was greater in the supine group versus the prone group (fig. 3B).
Lung Injury in Supine versus Prone Position
Overall lung injury was less in the prone versus supine groups in terms of wet/dry lung weight ratio (fig. 4A) and protein concentration in bronchoalveolar fluid (fig. 4B). The regional patterns of injury also differed between the groups. In the supine group, the lung tissue myeloperoxidase expression was higher in the dependent (dorsal) lung, in the same regions where spontaneous effort increased lung stress and inflation (67.5 ± 38.1 μm/min/mg protein vs. 167.7 ± 65.5 μm/min/mg protein in the nondependent vs. dependent lung; P = 0.003; fig. 5A), but there were no regional differences in myeloperoxidase expression in the prone group (61.0 ± 23.0 μm/min/mg protein vs. 74.0 ± 30.9 μm/min/mg protein in the nondependent vs. dependent lung; P = 0.951; fig. 5B). The distribution of “histologic” injury in each group is presented with illustrative sections (Fig. 5).
The prone position in severe ARDS has been traditionally used under passive conditions (i.e., under muscle paralysis or deep sedation).16 The current data suggest that the prone position could be an option to minimize lung injury from spontaneous effort in severe ARDS. This is because the prone position, independent of PEEP levels, diminishes the maldistribution of lung stress and thus the asymmetric, injurious lung inflation associated with spontaneous effort, and also because the prone position mitigates the magnitude of spontaneous efforts.
Ventilator-induced Lung Injury versus Effort-dependent Lung Injury
Using histology, computed tomography, and positron emission tomography imaging of [18F]fluoro-2-deoxy-d-glucose, previous studies revealed that ventilator-induced lung injury occurred in nondependent (ventral) lung regions in animal models of ARDS (rats, rabbits, pigs)6,22,23 and patients with ARDS.24,25 During a controlled breath, ventilation is likely to shift to nondependent (ventral) lung regions because of spatial heterogeneity of lung aeration, i.e., more atelectasis in the more dependent (dorsal) lung, and therefore a small percentage of the nondependent lung is more susceptible to higher inspiratory stress and strain in the supine position. On the other hand, the prone position decreases such spatial heterogeneity of lung aeration, leading to more even distribution of tidal strain and [18F]fluoro-2-deoxy-d-glucose uptake.26 Therefore, the prone position is known to reduce ventilator-induced lung injury12 and improve mortality in severe ARDS16
The current study confirmed that spontaneous effort altered the locus of lung injury: the bulk of effort-dependent lung injury occurred in the dependent (dorsal) lung,6,27,28 the same region where spontaneous inspiratory effort increased greater inspiratory lung stress and caused overinflation (figs. 2 and 5). Of note, the combination of low levels of PEEP and the supine position appears to pose the greatest risk of effort-dependent lung injury (fig. 2).
The maldistribution of lung stress during spontaneous effort was most manifested in the supine position with low PEEP. While the strength of spontaneous effort (measured as ΔPeso equals –10 cm H2O) was maintained to be the same among all conditions in the pig experiments, lower PEEP in the supine position was associated with the highest local lung stress in the dependent lung and thus the greatest magnitude of local lung inflation in the dependent lung. Therefore, overall lung injury from spontaneous effort was greater in the supine position (fig. 4), and the lung tissue myeloperoxidase expression and lung histologic injury were higher in dependent (dorsal) lung (fig. 5).
Mechanisms of Protection: Impact of Position
Prone position (vs. supine position) was effective to minimize effort-dependent lung injury, as evident from better gas exchange, better respiratory system compliance, lower wet/dry lung weight ratio, lower bronchoalveolar fluid protein concentration, and less lung tissue myeloperoxidase activity. The overall burden of lung injury was less, and there was no difference between degrees of injury in dorsal versus ventral lung. Several mechanisms were revealed from this study.
First, the prone position had no ventral to dorsal gradient in local Δ pleural pressure after diaphragmatic contraction, and therefore the magnitude of local lung inflation during spontaneous breathing is the same as under VT at approximately ≈7 ml/kg during muscle paralysis (fig. 2, C and D). This might be explained partially by the gravitational translocation of atelectatic solid-like lung tissue (which impedes pressure transmission) from the dorsal to ventral lung. This explanation is likely because “baby lung” is considered a functional entity (but not an anatomical entity) that can change its location with position and level of PEEP.11,29 Importantly, previous studies show that dorsal muscular regions of the diaphragm move more than ventral regions of the diaphragm during spontaneous breathing, regardless of the body position.30,31 Thus, the prone position decreases atelectatic solid-like lung tissue in the dorsal lung facing the well-moved, dorsal muscular regions of the diaphragm, which may facilitate the uniform transmission of Δ pleural pressure to the entire lung surface from where it was generated, after diaphragmatic contraction.
Second, the intensity of spontaneous effort (indicated by ΔPeso, respiratory rate in fig. 3A and table 1) was lower in the prone position (vs. supine position), despite matching levels of sedation. Because the prone position mitigates injurious spontaneous effort, it resulted in lower peak Δ transpulmonary pressure (i.e., dynamic lung stress). Of note, the benefit of the prone position in reducing the intensity of spontaneous effort was documented not only in rabbits (fig. 3A) but also in humans (infants,32,33 patients with ARDS,34 hypoxic patients with COVID-1935 ). Several plausible explanations are offered. First, the prone position increases end-expiratory lung volume in some patients (probably depending on lung recruitablity, the shape of the chest wall, the presence of abdominal hypertension, and the presence of support).11 In this study, the prone position was effective to recruit lung and increase lung volume in rabbits, suggested by higher respiratory system compliance in the prone position (table 1). Higher lung volume shortens diaphragm length, resulting in less force generation from the diaphragm.9,10 Second, the prone position per se is known to shorten diaphragm length even with the same end-expiratory lung volume as in the supine position, probably due to altered chest wall configuration and diaphragm geometry.36,37 Of course, the force generated by diaphragmatic contraction decreases as its length shortens.38
Therefore, the current study adds a promising technique to facilitate safe spontaneous breathing during mechanical ventilation in severe ARDS. It may synergize the benefits of spontaneous breathing (less muscle atrophy, more physiologic) with the benefits of the prone position per se (less ventilator-induced lung injury, more opening of well perfused regions). The current data support a larger clinical study as a next step to confirm the benefits of the prone position to render spontaneous effort less injurious in patients with ARDS whose spontaneous effort is vigorous.
Spontaneous Breathing and Prone Position Related to COVID-19
In the era of the COVID-19 pandemic, the indication of the prone position has been expanding: the prone position is now applied to nonintubated, hypoxic patients with COVID-19 (before intubation, not as severe as moderate-to-severe ARDS), hoping that being awake in the prone position might improve gas exchange, decrease the strength of spontaneous effort, minimizing the risk of effort-dependent lung injury, and thereby avoiding tracheal intubation.35,39,40 A few case reports observed that the prone position was associated with better gas exchange and lower respiratory rate,41,42 and a recent large randomized clinical trial has confirmed that being awake in the prone position significantly improved oxygenation, decreased the respiratory rate, and decreased the incidence of treatment failure and the need of intubation.35 Therefore, the beneficial effects of the prone position to mitigate effort-dependent lung injury has been found not only in mechanically ventilated patients with ARDS34 but also in nonintubated hypoxemic patients with COVID-19.35 The current physiologic study may reveal potential protective mechanisms of the prone position from spontaneous breathing.
There are several limitations to the current work. First, we utilized two different species (pigs and rabbits). Larger animals are closer to human physiology, so they are suitable for exploring the mechanism. The smaller animals are known to have a shorter (and steeper) trajectory of lung injury, so rabbits are more suitable for evaluating injury in a shorter time period; the overall consistent results in rabbits (location of injury) and pigs (location of lung stress and inflation) are reassuring. Second, different ventilatory modes were used (volume-controlled in pigs, pressure-controlled in rabbits). No differences in the patterns and magnitudes of dependent lung inflation imposed by spontaneous effort were observed between the volume-controlled mode and the pressure-controlled mode.7 Thus, the difference in ventilatory mode does not affect interpretation of the data. Third, the current study lacked paralyzed groups in the lung injury protocol. We chose supine and spontaneous breathing as a control group to compare with prone and spontaneous breathing. We cannot separate completely the benefits of lowering spontaneous effort from those of prone position per se. Fourth, our study included a single sex (males), the rationale being minimizing data variability. The potentially confounding effects of this sex bias on the meaning of single-sex experimental data should be considered.43
The current animal study found that the prone position, independent of PEEP levels, diminished a maldistribution of lung stress and thus asymmetric, injurious lung inflation associated with spontaneous effort and mitigated spontaneous effort, resulting in less effort-dependent lung injury.
Brian P. Kavanagh, M.D., sadly passed away on June 15, 2019. Prof. Kavanagh designed the study, interpreted all of the data, revised the manuscript, and was overall supervisor. The authors sincerely thank Prof. Kavanagh for his outstanding leadership and mentorship in completing this complex study.
Supported by Grant-in-Aid for Young Scientists 19K18294, funds from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Tokyo, Japan; to Dr. Yoshida), a Grant for the Promotion of Joint Research, funds from the Fukuda Foundation for Medical Technology of Japan (Tokyo, Japan; to Dr. Yoshida), a Research Training Competition (RESTRACOMP) Award, funds from the Hospital for Sick Children (Toronto, Ontario, Canada; to Dr. Yoshida), and funds from the Canadian Institutes of Health Research (Ottawa, Ontario, Canada; to Dr. Kavanagh).
The authors declare no competing interests.