Zero expiratory pressure and low oxygen concentration promote heterogeneity of regional ventilation and lung densities.

BACKGROUND: It is not well known what is the main mechanism causing lung heterogeneity in healthy lungs under mechanical ventilation. We aimed to investigate the mechanisms causing heterogeneity of regional ventilation and parenchymal densities in healthy lungs under anesthesia and mechanical ventilation.
METHODS: In a small animal model, synchrotron imaging was used to measure lung aeration and regional-specific ventilation (sV̇). Heterogeneity of ventilation was calculated as the coefficient of variation in sV̇ (CVsV̇ ). The coefficient of variation in lung densities (CVD ) was calculated for all lung tissue, and within hyperinflated, normally and poorly aerated areas. Three conditions were studied: zero end-expiratory pressure (ZEEP) and FI O2 0.21; ZEEP and FI O2 1.0; PEEP 12 cmH2 O and FI O2 1.0 (Open Lung-PEEP = OLP).
RESULTS: The mean tissue density at OLP was lower than ZEEP-1.0 and ZEEP-0.21. There were larger subregions with low sV̇ and poor aeration at ZEEP-0.21 than at OLP: 12.9 ± 9.0 vs. 0.6 ± 0.4% in the non-dependent level, and 17.5 ± 8.2 vs. 0.4 ± 0.1% in the dependent one (P = 0.041). The CVsV̇ of the total imaged lung at PEEP 12 cmH2 O was significantly lower than on ZEEP, regardless of FI O2 , indicating more heterogeneity of ventilation during ZEEP (0.23 ± 0.03 vs. 0.54 ± 0.37, P = 0.049). CVD changed over the different mechanical ventilation settings (P = 0.011); predominantly, CVD increased during ZEEP. The spatial distribution of the CVD calculated for the poorly aerated density category changed with the mechanical ventilation settings, increasing in the dependent level during ZEEP.
CONCLUSION: ZEEP together with low FI O2 promoted heterogeneity of ventilation and lung tissue densities, fostering a greater amount of airway closure and ventilation inhomogeneities in poorly aerated regions.

This study explored the forces that lead to heterogeneity in lung aeration. In this animal model with healthy lungs, zero airway pressure and lower inspired oxygen content contributed to inhomogeneous ventilation in more poorly aerated lung regions.

Worldwide, more than 230 million patients undergoing major surgery each year require general anesthesia and mechanical ventilation,1 and 20–30% of these patients are at intermediate to high risk for postoperative pulmonary complications,2 which affect clinical outcomes.3 Mechanical ventilation can initiate ventilator‐induced lung injury (VILI)4, 5 and contribute to extrapulmonary organ dysfunction.6 Up to 25% of patients with normal lungs when placed on mechanical ventilation for 2 days or longer will develop mild acute respiratory distress syndrome (ARDS), with 60–80% of those progressing to moderate to severe ARDS.5, 7, 8 A lung‐protective ventilation strategy in intermediate‐ and high‐risk patients undergoing major abdominal surgery improved clinical outcomes.9 A strategy to prevent VILI and ARDS is obviously desirable,5, 10, 11, 12, 13 but to achieve this, a better understanding of the primary mechanisms of VILI and ARDS development in healthy lungs under general anesthesia and mechanical ventilation is needed.

There is a growing recognition that ARDS following mechanical ventilation occurs often as a two‐hit phenomenon.11 A predisposing condition, such as an inflammatory endothelial (e.g. sepsis) or epithelial (e.g. aspiration) injury, is followed by a second insult that is predominantly mechanical in nature, resulting in activation of primed neutrophils and progression to ARDS. Strategies targeting risk factor modification, that is, avoiding a second‐hit in at‐risk patients have the potential to reduce rates of ARDS2, 11 and may guide in implementing lung‐protective ventilation strategies in patients without ARDS.3, 7, 14, 15, 16

A mechanical ventilation strategy that promoted stable lung recruitment and a more homogeneously ventilated lung, seems to have blocked early drivers of lung injury in normal lungs, preventing ARDS.10 The strategy comprised the use of airway pressure release ventilation with a prolonged time spent at the upper pressure that recruited alveoli gradually over time, resulting in a more homogeneously ventilated lung. A decrease in ventilation heterogeneity was also argued to be the main mediator of beneficial effects in other studies, reducing volutrauma and atelectrauma.5, 10, 11, 12, 13 But, it is not well known what is the main mechanism causing lung heterogeneity in healthy lungs under mechanical ventilation.

We hypothesized that low end‐expiratory pressure together with low oxygen concentration would promote heterogeneity of regional‐specific ventilation and lung tissue densities during mechanical ventilation. We therefore performed simultaneous measurements of parenchymal density, aeration, and regional‐specific ventilation, by means of synchrotron radiation computed tomography imaging, aiming to investigate the main mechanisms causing lung heterogeneity in healthy lungs. To fulfill this goal, we assessed the interplay between gravitational level, PEEP, and inspiratory oxygen fraction (FIO2) on heterogeneity of regional ventilation and lung tissue densities.

Methods

The procedures for the animal care and the experiments were in accordance with the Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes,17 and were approved by the Internal Evaluation Committee for Animal Welfare in Research of the European Synchrotron Radiation Facility, Grenoble, France.

Animal preparation

The experiments were performed on seven male New Zealand rabbits (2.8 ± 0.09 kg). A catheter (22 gauge) was inserted into the marginal ear vein under local anesthesia, using 5% topical lidocaine. Anesthesia was induced by the intravenous injection of 25 mg/kg of thiopental sodium. The animal was tracheotomized with a no. 3, Portex tube (Smiths Medical, Kent, United Kingdom). The left carotid artery and jugular vein were catheterized for blood gas measurements and for drug delivery, respectively. Anesthesia was then maintained with IV administration of 0.2 mg/kg/h of midazolam. After ensuring adequate anesthesia (no reaction by painful stimulation), continuous IV infusion of atracurium (1.0 mg/kg/h) was started. The animal was immobilized in the vertical, head‐up position in a cylindrical polyvinyl chloride custom‐made holder. Animals were in upright position throughout the study.

Pressure‐controlled mechanical ventilation was delivered using a custom‐made ventilator, described in detail previously.18 The ventilator allowed synchronizing mechanical ventilation with the image acquisition. Wash‐in image series of 15 images were recorded during inhalation of xenon (70%) and oxygen (30%), and ventilation was paused for 3 s for each image acquisition. The respiratory gas flow was monitored using a heated pneumotachometer (Hans Rudolph, Kansas City, MO, USA). The endotracheal pressure was monitored continuously. All monitored signals were amplified, digitized at 400 Hz (Powerlab, ADI Instruments, Oxfordshire, United Kingdom), and recorded on a computer.

Synchrotron radiation computed tomography imaging

The K‐edge subtraction (KES) method is a lung imaging technique that uses synchrotron radiation to quantify the regional concentration of inhaled stable xenon within lung airspaces.19, 20 This technique allows simultaneous and quantitative measurements of regional‐specific ventilation (sV̇) as well as lung tissue density. A detailed description of the methodology and instrumental setup has been extensively discussed in previous studies.19, 20, 21, 22 This imaging technique uses dual X‐ray beams at slightly different energies. X‐rays from a synchrotron radiation source are required because, as opposed to standard X‐ray sources, they allow the selection of monochromatic beams from the full X‐ray spectrum while conserving enough intensity for imaging with sufficient temporal resolution. Two computed tomography images are thus simultaneously acquired after the inhalation of a xenon–oxygen gas mixture. Visualization and quantitative measurement of xenon within the airways is based on the property that the attenuation coefficient of xenon increases by a factor of 5.4 when the energy of the incident X‐ray beam crosses the energy threshold of 34.56 keV, which is the xenon K‐edge energy. Using the dual‐energy KES imaging method, the densities due to tissue and xenon can be calculated separately in each voxel in the image, using a specifically developed computer algorithm19 explained in detail elsewhere.23 A ‘xenon‐density’ image allows the direct quantitative measurement of contrast gas within the airways, and that of the regional gas volume. Dynamic KES imaging during xenon wash‐in allows the measurement of regional sV̇.20 A ‘tissue‐density’ image obtained from the same data allows quantitative measurement of the regional tissue density.

The experiments were performed at the Biomedical Beamline of the European Synchrotron Radiation Facility (ESRF, Grenoble, France). The monoenergetic beams with an energy difference of 250 eV were produced from the continuous synchrotron radiation spectrum by a bent silicon crystal. The beams focused and crossed at the animal position, beyond which they diverged and were recorded by a liquid nitrogen‐cooled, high‐purity germanium, dual‐line detector (Eurisys Measure, Lingolsheim, France). The horizontal pixel size of the detector was 0.35 mm, and the vertical beam height was 0.7 mm. Image reconstruction was performed using the filtered‐back‐projection algorithm, using the Interactive Data Language (IDL; RSI, Boulogne‐ Billancourt, France).

Image analysis

Images were processed using the MatLab programming package (Mathworks Inc., Natick, MA, USA). Lung tissue was selected within the tissue‐density computed tomography images, by region‐growing segmentation. The volume of lung was then divided into four density categories: hyperinflated (comprising voxels with CT attenuation between: −1000 and −901 HU), normally aerated (between: −900 and −501 HU), poorly aerated (between: −500 and −101 HU), and non‐aerated tissue (between: −100 and +100 HU). A coefficient of variation in lung densities (CVD) was calculated for all of the lung tissue in the images, and within the normally aerated, hyperinflated, and poorly aerated lung regions.

The local specific ventilation or ventilation normalized to the gas volume within the voxel (sV̇) was calculated from the time constant of the xenon wash‐in using a single compartment model fit of xenon concentration vs. time.22 A 5 × 5 pixel moving average window was applied to the xenon‐density images prior to the model fit. The heterogeneity of ventilation was calculated as the coefficient of variation in sV̇ (CV) for all lung volume, and for areas within each density category contained in the image slice. In each sV̇ image, the histogram of sV̇ within normally aerated zones was calculated, and fit with a lognormal function. The median (μ) and standard deviation (σ) of the distribution were extracted from the fit. Normal, high, and low sV̇ were defined with reference to the median value of each slice. The ventilated alveolar volume was defined as the volume of the ventilation image where sV̇ was greater than the median of the distribution minus 2 standard deviations: μ−2σ.

Measurements were made at two axial levels, one close to the diaphragm and one 4.5 ± 1.9 mm (mean ± standard deviation) in the cranial, vertical direction. In order to characterize the functional behavior of normally aerated, poorly aerated, and hyperinflated lung regions, the volume of lung within each density category was further divided into subcategories based on regional sV̇: no ventilation, defined as: sV̇ < 0.5/min; low sV̇: 0.5/min < sV̇ < (μ−2σ); normal sV̇: sV̇ = μ ± 2σ; high sV̇: (μ + 2σ) < sV̇. Each subcategory was expressed as percentage of the total lung volume within the image slice.24

Study protocol

The horizontal X‐ray beam used for functional imaging in this study required that the animal be positioned vertically. Although the vertical posture may slightly improve respiratory mechanics in the rabbit,20 it has been already shown that it did allow assessment of the regional differences in lung function with respect to gravity.24 The following procedure was performed to eliminate the effects of the head‐up posture in terms of an increase in the end‐expiratory lung volume and respiratory system compliance: (1) Scanogram in horizontal posture and assessment and notation of the diaphragm position; (2) Measurement of gastric pressure in supine posture; (3) Position vertically in the custom‐made cylindrical polyvinyl chloride holder, exactly as during imaging; (4) Measurement of gastric pressure in vertical posture and inflation of inflatable cuff placed around the abdomen, in order to reach a gastric pressure equal to that in supine posture; (5) Check of diaphragmatic position; (6) Readjustment of cuff pressure if needed.

The animals were studied with the three following mechanical ventilation settings: PEEP = 12 cmH2O and FIO2 = 1.0 (OLP = ‘open lung PEEP’); PEEP = 0 cmH2O and FIO2 = 1.0 (ZEEP‐1.0); PEEP = 0 cmH2O and FIO2 = 0.21 (ZEEP‐0.21). The order of the three conditions was randomized (by random number), and upon each change in ventilator settings, 10 min were allowed for stabilization before imaging. A recruitment maneuver was performed before each step in the protocol in order to homogenize lung history. The settings of the recruitment maneuver were the following: pressure‐controlled ventilation, PEEP = 18 cmH2O, plateau pressure = 26 cmH2O, FIO2 = 1.0, respiratory rate set at 30, inspiratory to expiratory ratio (I:E) 1:1, during 4 min.

After the recruitment maneuver, tidal volume was 3 to 5 ml/kg and respiratory rate set at 50–60/min.

Statistical analysis

With a study design where the animals served as their own controls, four animals were needed to obtain a power of 0.80 with a P < 0.05, with SD = ½ mean. The time frame available for our experiments allowed us to use seven animals, giving us sufficient statistical power. The scatters in the parameters were expressed by the SEM values. The Shapiro–Wilk test was used to test data for normality. When data were normally distributed, within‐within‐subjects ANOVA with two within‐subject factors was applied. The two within‐subject factors were the mechanical ventilation settings (three conditions: OLP/ZEEP‐1.0/ZEEP‐0.21) and gravity level (two conditions: non‐dependent/dependent level). Where results showed significance, a Bonferroni adjustment for multiple tests was used for all pairwise comparisons. When the assumption of normality was violated, we used the Friedman test as non‐parametric alternative. Pairwise planned contrasts were then performed with a Bonferroni correction for multiple comparisons.

A one‐way repeated measures ANOVA was conducted to determine whether there were statistically significant differences in the gas exchange parameters over the three mechanical ventilation settings. The gas exchange data was normally distributed, as assessed by Shapiro–Wilk test (P > 0.05). Where results showed significance, a Bonferroni adjustment for multiple tests was used for all pairwise comparisons.

The statistical analyses were conducted by SPSS (version 20.0.0). Statistical tests were carried out with the significance level set at P value less than 0.05.

Results

The gas exchange parameters and the respiratory variables are shown in Table 1.

Table 1

Gas exchange parameters and respiratory variables

	OLP	ZEEP‐1.0	ZEEP‐0.21
pH	7.34 ± 0.1	7.22 ± 0.2	7.29 ± 0.1
PaO2/FIO2	423* ± 46.5	328.7 ± 131.3	227.9 ± 61.4
PaCO2	41.8 ± 8.0	71.7 ± 32.4	55.1 ± 14.8
HCO3 −	22.3 ± 3.1	26.2 ± 4.3	24.9 ± 2.1
VT	3.8 ± 0.8	3.6 ± 0.5	3.6 ± 0.5

Values presented are means ± SD. *P < 0.05 vs. ZEEP‐0.21. PEEP, positive end‐expiratory pressure; FIO2, inspired fraction of oxygen; OLP, PEEP 12 cmH2O and FIO2 1.0; ZEEP‐1.0, PEEP = 0 cmH2O and FIO2 1.0; ZEEP‐0.21, PEEP = 0 cmH2O and FIO2 0.21; pH, arterial pH; PaO2/FIO2 (mmHg), ratio of partial pressure of arterial oxygen to the fraction of inspired oxygen; PaCO2, arterial partial pressure of CO2; HCO3 −(mM), bicarbonate concentration; VT (mL/kg), tidal volume

At ZEEP, a heterogeneous distribution of mean density was seen at both horizontal image levels and the mean density increased from the non‐dependent to dependent level (P = 0.028). The mean tissue density under OLP was lower than ZEEP‐1.0 and ZEEP‐0.21 (P = 0.01 and 0.012, respectively; Fig. 1A), was more homogeneous, was within the normally aerated density category and similar between non‐dependent and dependent levels (Fig. 1A).

Figure 1

Panel (A) Mean tissue density. In the analysis of the mean tissue density, we found a significant main effect for mechanical ventilation settings (P = 0.003) and for gravity level (P = 0.028). In the pairwise comparisons with a Bonferroni adjustment for multiple tests, the mean tissue density of Open Lung‐PEEP (OLP) was significantly lower than Zero End‐Expiratory Pressure (ZEEP) ZEEP‐1.0 and ZEEP‐0.21 (* and †, P = 0.01 and 0.012, respectively). FiO2 = fractional inspired oxygen. HU = Hounsfield units. Panel (B) Coefficient of variation in lung densities for all . The coefficient of variation (CV) of lung densities calculated for all density categories together was significantly lower on Positive End‐Expiratory Pressure (PEEP) within both gravity levels (*, P = 0.011). Panel (C) Coefficient of variation in lung densities for poorly aerated density category. There was a statistically significant interaction between gravity level and mechanical ventilation settings on the coefficient of variation in lung densities calculated for the poorly aerated density category (P = 0.013). CV: coefficient of variation in lung densities; OLP: PEEP 12 cmH2O and F 2 1.0; ZEEP‐1.0: PEEP = 0 cmH2O and F 2 1.0; ZEEP‐0.21: PEEP = 0 cmH2O and F 2 0.21; white squares: nondependent level; black triangles: dependent level.

When calculated for all density categories together, the two gravity levels showed statistically significant changes in the CVD over the different mechanical ventilation settings (P = 0.011, Fig. 1B): predominantly, CVD increased during ZEEP.

There was a statistically significant interaction between gravity‐dependence and mechanical ventilation settings in the CVD calculated for the poorly aerated density category (P = 0.013, Fig. 1C), suggesting that the spatial distribution of the CVD calculated for the poorly aerated density category changed with the mechanical ventilation settings, increasing in the dependent level during ZEEP.

Compared to OLP, the area of lung regions with both low sV̇ and poor aeration was increased at ZEEP‐0.21 (P = 0.008, 12.9 ± 9.0 vs. 0.6 ± 0.4% for the non‐dependent level, and 17.5 ± 8.2 vs. 0.4 ± 0.1% for the dependent one). This is shown in Fig. 2.

Figure 2

Volume of lung with low regional‐specific ventilation within poorly aerated density category. In the analysis of the volume of lung with low sV̇ within the poorly aerated density category, a significant main effect for mechanical ventilation settings was found (P = 0.008). In the pairwise comparisons with a Bonferroni adjustment for multiple tests, the volume of lung with low sV̇ within the poorly aerated density category of ZEEP‐0.21 was greater than Open Lung‐PEEP (OLP) (*, P = 0.041). OLP: Positive End‐Expiratory Pressure (PEEP) 12 cmH2O and F 2 1.0; Zero End‐Expiratory Pressure (ZEEP) ZEEP‐1.0: PEEP = 0 cmH2O and F 2 1.0; ZEEP‐0.21: PEEP = 0 cmH2O and F 2 0.21; white squares: nondependent level; black triangles: dependent level; sV̇: regional‐specific ventilation.

Comparing the CV for total lung area between OLP vs. ZEEP‐0.21 and ZEEP‐1.0 pooled together, there was a statistically significant difference indicating more heterogeneity of ventilation during ZEEP irrespective of FIO2 (0.23 ± 0.03 vs. 0.54 ± 0.37, P = 0.049).

Figure 3 shows representative tissue density and sV̇ images at OLP, exemplifying its homogeneity of lung tissue density, and its preponderance of normally aerated units with normal sV̇. In contrast, it also shows representative images of ZEEP‐0.21, illustrating the emergence of a significant volume of lung with low sV̇ within the poorly aerated density category, i.e. with both low sV̇ (poorly ventilated) and poor aeration. It also shows images of ZEEP‐1.0.

Figure 3

Representative images of . Panel (A) This figure shows the homogeneity of lung tissue density, and the preponderance of normally aerated units with normal specific ventilation (sV̇) at the Open Lung‐PEEP (OLP) condition. And representative images of coincident poorly ventilation and poorly aeration during Zero End‐Expiratory Pressure (ZEEP) ZEEP‐0.21, showing the emergence of a significant volume of lung with low sV̇ within the poorly aerated density category, i.e. with both low sV̇ (poorly ventilated) and poor aeration. And representative images of ZEEP‐1.0. First row of images: Tissue‐density computed tomography image. Second row of images: Sample composite image showing the distribution of sV̇ (1/min). Third row of images: Volume of lung within each aeration category (normally aerated, poorly aerated, and hyperinflated lung regions) divided into subcategories: no ventilation [sV̇ < 0.5/min]; low sV̇ [0.5/min < sV̇ < (μ−2σ)]; normal sV̇ [sV̇ = μ ± 2σ]; high sV̇ [(μ + 2σ) < sV̇]. B) Each subcategory is expressed as percentage of the total lung volume within the image slice. The topographic distribution of lung regions defined as atelectatic, trapped, poorly aerated, normally aerated or hyperinflated is shown. Color scale indicates subregions with high, normal, or low sV̇ within poorly aerated (High D = high density), normally aerated (Normal D = normal density), and hyperinflated (Low D = low density) categories. Open Lung‐PEEP (OLP): Positive End‐Expiratory Pressure (PEEP) 12 cmH2O and F 2 1.0; Zero End‐Expiratory Pressure (ZEEP) ZEEP‐1.0: PEEP = 0 cmH2O and F 2 1.0; ZEEP‐0.21: PEEP = 0 cmH2O and F 2 0.21; sV̇ = regional‐specific ventilation. HU = Hounsfield units.

Discussion

We found that ZEEP together with low oxygen concentration increased the heterogeneity of both ventilation and lung tissue densities, likely fostering a greater amount of airway closure and ventilation inhomogeneities in poorly aerated lung regions. Conversely, OLP caused a more even distribution of aeration and of ventilation.

ZEEP‐0.21 resulted in larger regions of both poor aeration and ventilation compared to OLP, whereas ZEEP‐1.0 did not. The rate of absorption of gas from an unventilated region of the lung depends on the composition of the gas in that area when ventilation to it ceases, as well as the composition of the inspired gas.25, 26 When the inspired gas is 100% oxygen, the rate of collapse is faster than when air is breathed.25, 26 It has been demonstrated that if 100% oxygen is used, enough time of ‘stable air trapping’ is a prerequisite for lung collapse to occur.26, 27, 28, 29, 30, 31, 32 ‘Stable air trapping’ here means no cyclic opening and closing of the distal airways. If cyclic airway closure and reopening happen, no alveolar collapse would occur during a 10 min period unless the inspired ventilation–perfusion ratio (V AI/Q) is extremely low (<0.01), in practice, indistinguishable from continuous closure. We purposely used small tidal volumes to avoid as much as possible cyclical opening and closing (i.e. tidal recruitment) of the small airways that would enable ventilation of these distal lung units. If 21% oxygen, small tidal volumes and ZEEP are used altogether, ‘stable air trapping’ is a plausible consequence.

Low distending pressures and high FIO2 produced widespread dependent airway closure with subsequent absorption atelectasis in an experimental model of early ARDS, whereas low FIO2 substantially prevented atelectasis by much slower absorption of alveolar gas behind closed airways (‘stable air trapping’).28 Besides the fact that now we studied healthy lungs, here we add the relevant information that the latter condition also entails a greater heterogeneity in specific ventilation distribution.

Rylander and coworkers33 studied the size and distribution of the ‘poorly aerated’ density category in ARDS patients and concluded an uneven distribution of ventilation due to the presence of small‐airway closure and/or obstruction. The behavior of the ‘poorly aerated’ density category in our data expressed also an uneven distribution of ventilation likely mainly due to the presence of small‐airway closure and/or obstruction. Wellman et al.34 have shown that patterns of ventilation heterogeneity are compatible with airway narrowing or closure in poorly aerated lung regions. They found that increased ventilation heterogeneity during mechanical ventilation with high tidal volume and zero PEEP occurs primarily at length scales <60 mm, with a significant component derived from subresolution (<12 mm) length scales. Components of specific ventilation heterogeneity at length scales of <12, 12–36, and 36–60 mm were highest in poorly aerated regions.

Similar size alveoli have been shown to deform very heterogeneously under uniaxial stretch35 and even single cells deform differently due to variations in their stiffness.36 Although it is still unclear how the acinar and alveolar geometry would change the distribution of forces and strains in a heterogeneous acinus, ventilation of such structure is prone to interfacial stress injury. Heterogeneity of regional lung mechanical properties produces different energy storages between neighbor structures,37 which in turn may create damaging forces between neighboring units. This has an effect on regional ventilation and volume heterogeneity and would then be a possible cause of VILI. OLP can keep recruited lung regions and make stiffness more homogeneous at the length scales of small airways and alveoli,34 and hence reduce the risk of injury development.

In the heterogeneously expanding lung, the very initial triggering mechanism of local injury and inflammation may occur predominantly in regions likely to undergo either concentration of stress or cyclic recruitment. By using dynamic PET with [18F]‐Fluoro‐2‐deoxy‐D‐glucose in a porcine model of VILI, we recently showed that inflammation was higher within the normally and poorly aerated lung regions than in overdistended or collapsed regions.38 Besides the possibility of stress raisers at the intravoxel level (i.e. spots of accumulated tissue leveraging local stress), we can hypothesize other mechanisms explaining the marked inflammation in the poorly aerated density category: cyclic opening and closing of alveoli at the subvoxel level and cyclic opening and closing of small airways (also at subvoxel level) likely in poorly aerated regions with reduced specific ventilation, and increased tidal stretch, likely in normally aerated regions. The tidal opening and closing of distal bronchioles (‘bronchiolotrauma’ in analogy with the term atelectrauma) may be a key mechanism of VILI during mechanical ventilation. These findings reinforce and highlight the importance of better understanding of how ventilation heterogeneity is attenuated, exacerbated, and/or is modulated by the mechanical ventilation settings.

In an injured inhomogeneous lung, the stress and strain are unevenly distributed with localized increase in stress.39 The regional inhomogeneity may act as ‘stress raisers’.40 Cressoni et al. demonstrated that the lung inhomogeneities increased with ARDS severity, were positively associated with poorly aerated tissue, decreased when PEEP was increased, and were independently associated with outcome.40 The physiological dead space is likely one of the strongest predictors of the ARDS outcome because it is related to the structural changes of the lung.41 Cressoni et al. data exhibited that the only CT‐scan‐derived variable significantly associated with dead space fraction was the lung inhomogeneity extent. One can argue that this association between lung inhomogeneities with overall disease severity and mortality is meaningful only during ARDS and not in healthy lungs. However, the intensity of the lung inhomogeneities was similar in severe, moderate, and mild ARDS, and what actually changed between severe, moderate, and mild ARDS was only the extent of the lung inhomogeneities. The difference between patients with mild ARDS and patients with healthy lungs during major surgical procedures is likely also mainly a difference in the extent of the lung inhomogeneities. Another finding by Cressoni et al. was that the fraction of the poorly aerated tissue was well associated with the extent of inhomogeneity.

This study had several limitations. The technical aspects of analyzing structural and functional changes using KES imaging have been extensively discussed previously.20 Given a potential difference in end‐expiratory lung volume between PEEP of 12 cmH2O and ZEEP and due to the fact that only two slices were analyzed, we cannot exclude that we analyzed different anatomical areas. The horizontal X‐ray beam used for functional imaging in this study required that the animal be positioned vertically. This posture may slightly improve the respiratory mechanics in rabbit.20 However, in this study, we compensated for this effect by raising the external pressure around the abdomen. Finally, this investigation was done in a small animal model.

In conclusion, in healthy lungs under mechanical ventilation, the present findings suggest that ZEEP together with low oxygen concentration promotes heterogeneity of ventilation and that of lung tissue densities, likely fostering a relevant amount of airway closure and ventilation inhomogeneities in poorly aerated lung regions. Furthermore, such heterogeneities can be minimized by OLP. The present data denote increased impedance of the subtending airways conducting gas to poorly aerated regions, either through narrowing or due to closure of these small airways, which may play a relevant role as potential early drivers of VILI.