Individualized Positive End-expiratory Pressure and Regional Gas Exchange in Porcine Lung Injury

Although positive pressure ventilation is routinely used in critically ill patients with respiratory failure or acute respiratory distress syndrome (ARDS), mechanical ventilation may aggravate lung injury by mechanical stress transferred to lung tissue.¹  Reduction in tidal volumes²  (V_T), but not high positive end-expiratory pressure (PEEP), has been shown to improve survival in ARDS patients.³  Clinicians tend to increase inspiratory oxygen rather than PEEP⁴  and do not consistently follow recommended strategies to adjust PEEP.⁴  How to best set PEEP has been hashed out for decades,⁵  and yet not a single prospective study has been able to provide a definitive answer.

Lower PEEP/no recruitment strategies aim at finding the lowest possible PEEP that ensures an acceptable oxygenation while limiting plateau pressures and inspiratory overdistension,²  and tolerating lung collapse (permissive atelectasis).^2,6  Recent clinical guidelines do not recommend the routine use of recruitment maneuvers.^7–9  Derecruitment, however, may promote opening and closure of small airways and alveoli (tidal recruitment)^10,11  and intrapulmonary shunting.

Higher PEEP/full recruitment strategies following a recruitment maneuver aim to optimize oxygenation as surrogate for full lung recruitment,^12,13  and to minimize tidal recruitment, and atelectrauma.¹⁰  However, high PEEP levels may cause high end-inspiratory airway pressure, which might impair regional perfusion.^5,14,15 

Elevated intraabdominal pressure aggravates lung collapse, tidal recruitment, inhomogeneity of regional ventilation, ventilation ()/perfusion () mismatch,¹⁶  and is frequently observed in mixed populations of mechanically ventilated patients,^16,17  especially if lung failure or ARDS is caused by extrapulmonary reasons.^17,18  In patients with potential for lung recruitment, individualized PEEP setting to detect and improve lung recruitment¹¹  and to minimize tidal recruitment¹⁰  may improve outcomes.

Electrical impedance tomography noninvasively allows bedside monitoring of regional ventilation^19,20  and can be used for individual PEEP titration.^19–21  Electrical impedance tomography–based quantification of inhomogeneity in regional ventilation time courses (regional ventilation delay inhomogeneity)^22,23  has been shown to correlate linearly with tidal recruitment.^22,23  Low tidal recruitment indicates sufficient lung recruitment and may thus enhance homogeneity of ventilation () and perfusion () and, thus, reduce mismatch at the lowest possible PEEP level.

We hypothesized that an electrical impedance tomography–based approach minimizing inhomogeneity of regional ventilation time courses as measure of tidal recruitment improves matching when compared to either low PEEP or high PEEP strategies. This hypothesis was studied using single photon emission computed tomography in a porcine hybrid model combining recruitable lung injury with elevated intraabdominal pressure.

Fifteen healthy pigs (nine males, six females) were anesthetized and mechanically ventilated in supine position. Anesthesia^22,24  was induced with intramuscular atropine (0.04 mg/kg), tiletamin–zolazepam (6 mg/kg), and xylazine (2.2 mg/kg) and maintained by continuous infusion of ketamine (10 mg · kg^-1 · h^-1), midazolam (0.4 mg · kg^-1 · h^-1), fentanyl (10 µg · kg^-1 · h^-1). Depth of anesthesia was verified by paw pinch before animals were continuously paralyzed using pancuronium bromide (0.15 mg · kg^-1 · h^-1) and absence of spontaneous breathing activity was confirmed by observation of continuously displayed gas flow tracing.^22,24  Tracheotomy and instrumentation were performed as previously described.^22,24  To induce lung injury, abdominal pressure (measured in the urinary bladder) was increased to 15 mmHg by infusion of 0.9% saline into the abdominal cavity,^22,24,25  followed by titrated central venous injections of oleic acid, until a stable Pao₂/fractional inspired oxygen concentration (Fio₂) ratio of less than 200 mmHg was reached.^22,24,26 

Heart rate and systemic, central venous, and pulmonary artery blood pressures were measured using arterial, central venous, and pulmonary arterial cannulas.^22,24  Cardiac output, extravascular lung water, and intrathoracic blood volume were determined using transpulmonary thermodilution.^22,24  Systemic vascular resistance was calculated using standard equations.

Waveforms were measured using the integrated respiratory monitor of the ventilator (Engström Carestation; GE Healthcare, Germany) and stored for offline analysis.²²  Intraabdominal pressure was measured intermittently in the urinary bladder as described previously.^22,24  Blood gases, oxygen saturation, and hemoglobin levels were measured using a cooximeter (Radiometer, Germany). Venous admixture was calculated using standard equations.²⁷ 

Regional ventilation delay inhomogeneity was measured during a slow inflation (12 ml/kg ideal body weight) using electrical impedance tomography (EIT evaluation KIT II; Dräger Medical, Germany)^22,23  to estimate the amount of tidal recruitment by measuring inhomogeneity in regional ventilatory time courses (fig. 1 and Supplemental Digital Content 1, http://links.lww.com/ALN/C206).

Spatial ventilation (81^mKrypton gas) and perfusion (99^mTecnetium-labeled macroaggregated albumin) distributions^28,29  were analyzed during ventilation with different PEEP levels. Images were acquired on a dual-head gamma camera and reconstructed after filtering, noise and background correction.³⁰  For each voxel we calculated:

• regional ventilation/voxel ([] gas flow per voxel)

• regional perfusion/voxel ([] blood flow per voxel)

• ratio per voxel³⁰ 

All voxels were assigned to one of the following compartments according to their ratio (using thresholds known from multiple inert gas elimination technique³¹ ) (fig. S2, Supplemental Digital Content 2, http://links.lww.com/ALN/C207):

• shunt ( < 0.005)

• low (0.005 ≤ < 0.1)

• normal (0.1 ≤ < 10)

• high (10 ≤ < 100)

• dead space ( ≥ 100)

Lung tissue volume, perfusion, and ventilation of compartments were calculated (fig. S2, Supplemental Digital Content 2, http://links.lww.com/ALN/C207):

• shunt compartment: lung tissue volume undergoing shunt perfusion

• shunt perfusion: amount of blood flow that is distributed to the shunt compartment

• ventilated lung: all compartments other than shunt compartment

• dead space compartment: lung tissue volume undergoing dead space ventilation

• dead space ventilation: amount of gas flow that is distributed to the dead space compartment

• perfused lung: all compartments other than dead space compartment

Whereas single photon emission computed tomography captured the whole lung and yielded voxels of 4.42 × 4.42 × 4.2 mm (= 0.0821 ml/voxel) electrical impedance tomography analyzed a representative lens-shaped region of approximately 18.5 × 18.5 × 9.25 cm reconstructed to 32 × 32 pixels of approximately 5.8 ×.5.8 mm (fig. S3, Supplemental Digital Content 3, http://links.lww.com/ALN/C208).

Average lung aeration was measured by low resolution density scans (transmission scans from single photon emission tomography during ongoing ventilation). The lung was manually marked (Osirix v. 5.5, Switzerland) in ten equidistant (cranio-caudal) slices³²  and lung volumes and masses were calculated in differently aerated lung compartments (nonaerated, poorly aerated, normally aerated, overaerated)^22,33  as percentage of total lung volume and total lung mass (Supplemental Digital Content 4, http://links.lww.com/ALN/C209). During all offline image analyses the investigator was blinded to the PEEP.

Volume-controlled mechanical ventilation was applied with a V_T of 6 to 8 ml/kg, an inspiratory-to-expiratory ratio of 1:1, an Fio₂ of 0.5, and a PEEP of 5 cm H₂O. Respiratory rate was 25 to 30 breaths/min.²² 

After induction of lung injury, respiratory rate had to be increased to 30 to 40 breaths/min to avoid severe hypercapnic acidosis (Paco₂ greater than 60 mmHg; pH less than 7.25), while keeping inspiratory-to-expiratory ratio constant and ensuring complete expiration as observed by a zero end-expiratory flow.²²  Spontaneous breathing efforts were continuously suppressed by muscle paralysis, which was confirmed by absence of spontaneous breathing activity in the continuously displayed airway pressure and flow tracings.

Individualized PEEP levels were guided by three different strategies to be compared later in a cross-over design (fig. 2 and Supplemental Digital Content 5, http://links.lww.com/ALN/C210).

PEEP and Fio₂ were adjusted according to the ARDS Network protocols lower PEEP/ Fio₂ table²  without a preceding recruitment maneuver²  aiming at lung rest and tolerating “permissive atelectasis.” Resulting PEEP was defined as “table PEEP.”

A lung recruitment maneuver¹³  was performed targeting a Pao₂ greater than 400 mmHg at Fio₂ = 1.0 while using pressure-controlled ventilation with a driving pressure of 15 cm H₂O and increasing PEEP from 30 up to 45 cm H₂O in steps of 5 cm H₂O every 2 min. Then, PEEP titration was performed using volume-controlled ventilation (6 to 8 ml/kg ideal body weight) starting at 30 cm H₂O. Every 4 min blood gases were taken and tidal recruitment²²  was measured during a single low flow breath using electrical impedance tomography (fig. 2), as previously described, before PEEP was decreased in steps of 2 cm H₂O.

Deterioration of oxygenation was defined as Pao₂ decrease of more than 5% from the individual maximum with decreasing PEEP. “Maximal oxygenation PEEP” was defined as the lowest PEEP level that avoided this decrease and was set 2 cm H₂O above the PEEP that caused a Pao₂ decrease (fig. 1). Deterioration of temporal homogeneity was defined as increase in regional ventilation delay inhomogeneity with decreasing PEEP. “Minimal tidal recruitment PEEP” was defined as the lowest PEEP level that avoids this increase and was set 2 cm H₂O above the PEEP that caused an increase in regional ventilation delay inhomogeneity (fig. 1).

After transfer to the scanner laboratory without interrupting mechanical ventilation using baseline settings, table PEEP, maximal oxygenation PEEP, and minimal tidal recruitment PEEP (based on previous titration) were studied in randomized order (blockwise randomization, blocks of six, sealed envelopes), while keeping all other ventilator settings constant.

Before every measurement point, stability of lung injury was checked, and carry-over effects of lung volume history were prevented by following standard sequence (fig. 2):

• Table PEEP: derecruitment (disconnection from the ventilator, PEEP 5 for 1 min), stability check (table PEEP for 10 min, blood gas analysis), complete sets of measurements including single photon emission computed tomography after additional 20 min of ventilation with table PEEP.

• Maximal oxygenation PEEP: derecruitment and stability check as described above (table PEEP for 10 min, blood gas analysis), recruitment maneuver using the previously used pressures (during maximal recruitment in the lab), complete sets of measurements including single photon emission computed tomography after 20 min of ventilation with maximal oxygenation PEEP.

• Minimal tidal recruitment PEEP: derecruitment and stability check as described above (table PEEP for 10 min, blood gas analysis), recruitment as described above, complete sets of measurements including single photon emission computed tomography after 20 min of ventilation with minimal tidal recruitment PEEP.

After approval by the local Animal Research Ethics Committee (approval No. C274/7), this study was performed in the Hedenstierna Laboratory, Department of Clinical Physiology, Uppsala University Hospital, Uppsala, Sweden, in adherence with the Guide for the Care and Use of Laboratory Animals (National Academy of Science 1996).

For this exploratory study, no reliable pilot data or data from publications were available. Based on experiences with other animal model studies using different ventilatory settings,^22,24  we assumed differences between levels in the range of 75% of the SD at each level and a correlation of 0.5 between levels. A target power of 80% was used. With a sample size of 15, the test of a single contrast between two PEEP settings at a 0.05 α level in a one-way repeated measures ANOVA with three levels was found to have 80.3% power to detect a contrast C of 3, assuming SD = 4 at each level, a between-level correlation of 0.5, and an a resulting effect size of 0.752. We followed the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines³⁴  (Supplemental Digital Content 6, http://links.lww.com/ALN/C211). Primary outcome measures were differences in PEEP levels and amounts of gas and blood flow to different compartments.

Data (expressed as mean ± SD) were tested for normal distribution (Shapiro–Wilk test) and analyzed using two-tailed testing. Paired t tests, one- or two-way repeated-measures ANOVA) and consecutive post hoc tests (Newman–Keuls, repeated-measures ANOVA), were performed if appropriate. Results from blood gas analyses, shunt, and atelectasis were compared using linear correlation (Pearson) and Bland–Altman analysis. (STATISTICA for Windows 6.0; StatSoft, Inc., USA). P < 0.05 was considered to be statistically significant (for details see Supplemental Digital Content 7, http://links.lww.com/ALN/C212).

All animals finished the whole study protocol. Due to invalid raw data, one pig (No. 1) had to be excluded. No outlier data were excluded. Finally, 14 animals were analyzed.

Our lung injury model caused in all animals an oxygenation and lung mechanics impairment compatible with the current criteria for moderate human ARDS³⁵  (table S1, Supplemental Digital Content 8, http://links.lww.com/ALN/C213) and was stable throughout the entire study period (table S2, Supplemental Digital Content 9, http://links.lww.com/ALN/C214). Randomization resulted in homogeneous distribution of all orders of measurements (table S3, Supplemental Digital Content 10, http://links.lww.com/ALN/C215).

Table PEEP resulted in the lowest PEEP (table PEEP, 11 ± 3; minimal tidal recruitment PEEP, 22 ± 3; maximal oxygenation PEEP, 25 ± 4 cm H₂O; P < 0.001; table 1) and plateau (33 ± 6 vs. 37 ± 6 vs. 41 ± 7 cm H₂O; P < 0.001; table 1), but highest driving pressure (22 ± 6 vs. 15 ± 4 vs. 16 ± 6 cm H₂O; P < 0.001; table 1), whereas PEEP and plateau pressure were highest with maximal oxygenation PEEP. Driving pressure was lower using both minimal tidal recruitment and the maximal oxygenation PEEP. As indicated by a decrease in regional ventilation delay inhomogeneity measured by electrical impedance tomography, minimal tidal recruitment PEEP reduced tidal recruitment, when compared to table PEEP. Of note, further increase of PEEP using maximal oxygenation PEEP did not further decrease regional ventilation delay inhomogeneity (table PEEP, 9.0 ± 3.6; minimal tidal recruitment PEEP, 4.8 ± 1.3; maximal oxygenation PEEP, 5.6 ± 2.2%; P < 0.001; table 1).

Higher PEEP levels during minimal tidal recruitment PEEP and maximal oxygenation PEEP improved venous admixture (table 1) and arterial oxygenation (table PEEP, 85 ± 20; minimal tidal recruitment PEEP, 257 ± 94; maximal oxygenation PEEP, 240 ± 100 mmHg; P < 0.001; table 1). Intrathoracic blood volume was comparable between all PEEP settings (table S4, Supplemental Digital Content 11, http://links.lww.com/ALN/C216).

During PEEP titration, which lasted 4 min at any PEEP level, maximal oxygenation PEEP, as intended, resulted in the highest Pao₂ when compared to minimal tidal recruitment PEEP (483 ± 126 vs. 441 ± 125 mmHg; P = 0.0192; paired t test). These short-term differences were not sustained when respective PEEP levels were used for a longer period (20 min) during single photon emission tomography scans (table 1) which may be explained by extrapulmonary long-term effects such as lower cardiac output at highest PEEP levels (table 1).

Pulmonary gas volume was comparable between minimal tidal recruitment and maximal oxygenation PEEP (table 2). In contrast, the table PEEP, which was not preceded by a recruitment maneuver, resulted in a reduced gas volume (table 2). Total lung mass was comparable between all three PEEP settings (table 2). (For detailed results of quantitative computed tomography analyses see figs. S4 to S6, Supplemental Digital Content 12, http://links.lww.com/ALN/C217.)

Figure 3 shows spatial distribution of regional ventilation and perfusion. Volumes, ventilation, and perfusion of lung compartments are given in figure 4. Ventilated lung volume was reduced with table PEEP (fig. 3A) and normal compartment amounted to about 25% of total lung (fig. 4A). In contrast, normal compartment was more than doubled with minimal tidal recruitment and maximal oxygenation PEEP (fig. 4A). Using table PEEP, the largest lung proportion was perfused but not ventilated (shunt compartment, fig. 4A). In contrast, ventilation was redistributed to the dorsal lung regions (fig. 3D) and the shunt compartment decreased during both minimal tidal recruitment and maximal oxygenation PEEP (fig. 4A). The volume of ventilated, but not perfused, lung tissue (dead space compartment) was largest using maximal oxygenation PEEP (fig. 4A).

During table PEEP, the perfusion of normal compartment was lower when compared to both other strategies, whereas perfusion of shunt compartment was tripled and amounted to more than 50% of cardiac output (fig. 4B). With higher PEEP during both minimal tidal recruitment and maximal oxygenation PEEP, overall pulmonary perfusion was reduced (figs. 3B and 4B) mainly reflected by reduction in shunt perfusion (fig. 4B) in dependent lung regions (fig. 3C). Whereas absolute blood flows were different (fig. 3B), relative regional perfusion distribution along the normalized ventral-to-dorsal axis was not affected by these PEEP strategies (fig. 3B).

On the normalized ventral to dorsal axis, ventilation was redistributed from nondependent to dependent lung regions during both minimal tidal recruitment and maximal oxygenation PEEP when compared to table PEEP (figs. 3D and 5, second row). Ventilation of the normal compartment was reduced with table PEEP (fig. 4C), and dead space ventilation (gas flow to the dead space compartment) in the ventral lung regions was nearly doubled when compared to both other PEEP strategies (figs. 3E and 4C) and contributed to more than one third of the total minute ventilation.

With decreasing aerated lung volume during table PEEP, regional ventilation (gas flow per voxel) of the remaining aerated voxels reached up to 2 ml · min^-1 · voxel^-1, primarily in the more ventral lung regions, whereas regional ventilation remained less than 1 ml · min^-1 · voxel^-1 in all lung regions using the minimal tidal recruitment and maximal oxygenation PEEP (figs. 3F and 5, second row).

Figure 5 shows two- and three-dimensional reconstructions of regional ventilation and perfusion analyses from a representative animal for the three PEEP strategies (of note: this illustrative example is not necessarily representative for all quantitative results of the whole study population). Comparison of PEEP titrations based on electrical impedance tomography and driving pressure are given in the online supplement (figs. S7 and S8; table S5; Supplemental Digital Content 13, http://links.lww.com/ALN/C218). A comparison of regional measurements and blood gas analyses is provided in the online supplement (figs. S9 to S11, Supplemental Digital Content 14, http://links.lww.com/ALN/C219).

This animal study comparing three strategies for individualized PEEP titration in a porcine model of recruitable lung injury and elevated intraabdominal pressure showed that different strategies resulted in different PEEP levels. Minimal tidal recruitment and maximal oxygenation PEEP caused and maintained alveolar recruitment, decreased regional ventilation/voxel, reduced both shunt compartment and shunt perfusion, and diminished dead space ventilation, when compared to table PEEP. Dead space compartment was highest with maximal oxygenation PEEP.

Single photon emission tomography enables discrimination of shunt perfusion (blood flow to nonventilated lung volume) from the size of shunt compartment (volume of shunt-perfused lung tissue). The high amounts of both shunt compartment and shunt perfusion during table PEEP were reduced with higher airway pressures during both other PEEP settings. Two different (recruitment-dependent and recruitment-independent) but concordant mechanisms can explain this finding. First, alveolar recruitment decreased the size of the shunt compartment (figs. 4A and 5) and, hence the corresponding blood flow (fig. 3B) no longer appeared as shunt perfusion (fig. 3C). Second, as previously shown, a decreased cardiac output resulting from reduced cardiac preload³⁶  with higher intrathoracic pressures may cause a disproportionately higher reduction in shunt perfusion,³⁷  which is in agreement with our data (fig. 3, B and C). In contrast to experimental data from a lavage model,³⁸  we did not find redistribution of pulmonary blood flow toward dorsal regions with increase in PEEP. This may be explained by an already increased perfusion of these regions due to attenuated hypoxic pulmonary vasoconstriction in our model.³⁹ 

In analogy, regional analyses allow discrimination of the size of the dead space compartment (volume of ventilated but nonperfused lung tissue) from dead space ventilation (gas flow to this dead space compartment). Dead space, as defined here, is where ventilation exceeds perfusion by the factor 100.^31,40  This implies that the dead space compartment, or part of it, can be ventilated by a very small gas flow as long as it exceeds blood flow by the factor 100. Depending on the presence of high or low gas flows, the same amount of dead space ventilation can either be caused by low regional ventilation/voxel (fig. 3F) of large dead space compartment (implicating a static distension of aerated lung tissue) or by high regional ventilation/voxel (fig. 3F) of small dead space compartment (implicating a more dynamic intratidal distension of aerated lung tissue). Following the aforementioned considerations, impairment of oxygenation and carbon dioxide elimination is rather affected by the amount of shunt perfusion and dead space ventilation than by the sizes of the shunt and dead space compartments.

Since V_T, minute ventilation, and respiratory rate remained constant between the different PEEP strategies, changes in regional ventilation/voxel (fig. 3F) were mainly a function of differences in ventilated lung volume (fig. 3A). Aeration data suggest that differences in lung volume were mainly caused by recruitment of collapsed lung rather than by changes in edema because lung masses were similar between PEEP settings (figs. S5 and S6, Supplemental Digital Content 12, http://links.lww.com/ALN/C217).

At lowest airway pressures during table PEEP without a preceding recruitment maneuver, the derecruited, small lung (table 2; fig. 3A) had to accommodate the total V_T, which resulted in higher regional ventilation/voxel (fig. 3F). This suggests increased specific ventilation (volume change/resting volume) related to strain,^41–43  which causes ventilator-induced lung injury.^41,44  Increased driving pressure is associated with poor outcome in ARDS patients.^4,45  Highest driving pressures were observed with table PEEP due to lowest respiratory system compliance and lung volume (“baby lung”)⁴⁶  promoting cyclic inspiratory distention⁵  even with “protective” V_T. This is supported by previous findings in ARDS patients⁴⁷  and data from a ventilator-induced lung injury model showing inflammation⁴⁸  of ventilated ventral lung regions, whereas collapsed regions are “protected” by being non- or poorly-ventilated.⁴⁸  In our study, increased regional ventilation/voxel during table PEEP (fig. 3F) was associated with dead space ventilation (fig. 3E), which causes mechanical stress without contributing to gas exchange (wasted ventilation).⁴⁹  Increased dead space ventilation has been shown to be associated with impaired outcome.⁵⁰ 

In our model characterized by a recruitable lung injury, both high PEEP strategies, minimal tidal recruitment PEEP and maximal oxygenation PEEP, were comparably associated with sustained alveolar recruitment (table 2) increasing ventilated lung volume (fig. 3A) and respiratory system compliance while reducing driving pressures. However, the preceding recruitment maneuver may at least partially explain these results. Redistribution of ventilation to recruited lung regions caused reduction in regional ventilation/voxel of ventral lung regions (fig. 3F) which might decrease cyclic strain. Dead space compartment (fig. 4A) increased with higher PEEP levels and was highest with maximal oxygenation PEEP. Whether this is attributable to static strain cannot be derived from our data, since we did not measure lung distension. Finally, minimal tidal recruitment PEEP ensured comparable lung recruitment and volume, regional ventilation/voxel, and regional gas exchange at lower PEEP levels when compared to the maximal oxygenation PEEP strategy, but requires additional technical equipment which makes this impedance tomography–based strategy less clinically compelling.

Clinical data suggest increased mortality in patients with high amount of collapsed but recruitable lung tissue undergoing tidal recruitment.¹¹  Decreased mortality has been associated with lung recruitment and reduced tidal recruitment at higher PEEP levels,^3,10  whereas patients with less recruitable lungs may face harm from inspiratory overdistension.¹⁰  Although increasing airway pressures to facilitate lung recruitment before PEEP titration seems physiologically reasonable—at least in patients with recruitment potential—recruitment maneuvers with pressures up to 60 cm H₂O followed by PEEP titration aimed at the highest global compliance in patients with mainly pneumonia-associated, less recruitable ARDS increased mortality,⁵¹  and recent clinical guidelines do not recommend the routine use of recruitment maneuvers. In patients with less recruitable lungs (e.g., due to pneumonia), application of high PEEP and airway pressures to facilitate recruitment as performed in our study maybe harmful. Hence, identification of recruitability is crucial.

PEEP setting according to respiratory system or lung mechanics (e.g., by using esophageal pressure-guided approaches) improved oxygenation,⁵²  but not survival.⁵³  Minimizing temporal lung inhomogeneity guided by impedance tomography may not individually result in the same PEEP levels as found by using measures of global lung mechanics (Supplemental Digital Content 13, http://links.lww.com/ALN/C218), suggesting that global and regional information on lung function might differ. PEEP titration was also performed in different animal models and humans measuring lung volume, ventilation distribution, and regional lung mechanics by impedance tomography.¹⁹  Whether impedance tomography–based strategies are clinically advantageous warrants further investigations.

Our study has limitations. Firstly, acute lung failure and ARDS summarize several pathophysiologic entities, which cannot be mimicked by a single experimental model. Elevated intraabdominal pressure is frequently observed in ventilated patients.^16–18  Whereas its incidence in all ARDS patients is unknown, elevated intraabdominal pressure is prevalently seen in respiratory failure or ARDS caused by extrapulmonary reason^16–18  (e.g., abdominal sepsis).

We used intraabdominal saline infusion to increase intraabdominal pressure as previously reported.^22,24,54  Others used air insufflation²⁵  or intraabdominal balloons.^26,55  We combined oleic acid injection,⁵⁶  which causes endothelial lung injury,⁵⁶  with moderately elevated intraabdominal pressure,²⁵  as done before.^22,24,26,55  Adding elevated intraabdominal pressure^22,24,26  aggravates experimental lung injury²⁶  and increases collapse and tidal recruitment.^22,24  Thus, our hybrid model allows stable lung collapse and recruitability over time. However, our results are not necessarily transferrable to all ventilated patients, and especially not to pulmonary-induced ARDS characterized by less recruitability. Since intraabdominal pressure affects transpulmonary pressure, our model might favor higher PEEP levels and results might differ with other intraabdominal pressure levels. However, comparable effects on recruitment and global gas exchange were recently reported when applying these PEEP strategies in aspiration-induced lung injury without elevated intraabdominal pressure.²¹  Moreover, PEEP according to the PEEP/Fio₂ table is frequently used as a reference strategy when studying different PEEP strategies in patients with^57,58  and without^59–61  elevated intraabdominal pressure.

Second, due to the lack of a recruitment maneuver, the table PEEP strategy started from a different volume history (derecruited lung) than both other strategies (fully recruited lung). Using the ARDS Network higher PEEP/Fio₂ table, and/or using table PEEP after a recruitment maneuver would likely result in lung recruitment and less mismatch. Thus, by design, our study is biased in this regard, and we cannot differentiate between effects of a recruitment maneuver and high PEEP. We compared clinically relevant strategies aiming at lung recruitment or lung rest and “permissive” atelectasis. The ARDS Network studies neither proposed recruitment maneuvers nor excluded patients with elevated intraabdominal pressure.² 

Third, the impedance tomography–based method of PEEP titration was previously validated in different lung injury models characterized by low to high recruitability.^21,22,24  In animals without lung injury, PEEP titration resulted in low regional ventilation delay inhomogeneity already at low PEEP levels providing evidence that the impedance tomography–based method is sensitive in nonrecruiters as well (fig. S1, Supplemental Digital Content 1, http://links.lww.com/ALN/C206). Although increased inhomogeneity values at high PEEP levels (fig. 1) can be explained by delayed inflation of distended lung tissue,²²  the method is not validated to detect alveolar hyperdistention.

Fourth, different arbitrary thresholds of oxygen decline have been reported (Pao₂, 10% decrease; less than 450 to 400 mmHg)^13,62  to ensure “optimal gas exchange.”¹²  We used a 5% decrease in Pao₂ to assume maximum recruitment as an upper extreme. Using other oxygenation thresholds, the maximal oxygenation PEEP level might have been different.

Fifth, results for dead space may have been different using pressure-controlled ventilation. Additionally, by design, we could not study effects of PEEP strategies on ventilation-induced lung injury and outcome.

Last, shunt values from regional measurement correlated well to shunt values measured by blood gas analysis and atelectasis. However, we found less accuracy with higher shunt values which can be explained by basically differences between both methods (see Supplemental Digital Content 14, http://links.lww.com/ALN/C219). We cannot validate dead space ventilation from our regional data. Single photon emission computed tomography does not enable detection of all anatomical dead space. However, the gross appearance of the distribution is similar to that of the classic multiple inert gas elimination technique.⁴⁰  Although dead space ventilation and dead space compartments might suggest dynamic or static hyperinflation of the lung, our densitometric analysis did not show hyperaerated lung tissue. This might be explained by interaction between density thresholds, slice thickness, and reconstruction parameters impairing the detection of hyperinflation.⁶³ 

In a pig model of recruitable lung injury and elevated intraabdominal pressure, three PEEP strategies resulted in different PEEP levels. When compared to table PEEP without a recruitment maneuver, both minimal tidal recruitment PEEP and maximal oxygenation PEEP following a recruitment maneuver induced lung recruitment, which decreased shunt perfusion, dead space ventilation, and regional ventilation/voxel. Effects of minimal tidal recruitment PEEP and maximal oxygenation PEEP were comparable, but with lower PEEP in the minimal tidal recruitment PEEP setting.

The authors thank Eva-Maria Hedin (research engineer, Department of Clinical Physiology, University of Uppsala, Uppsala, Sweden), Anne Abrahamson (research assistant, Department of Clinical Physiology, University of Uppsala), and Agneta Roneus (senior research engineer, Department of Clinical Physiology, University of Uppsala) for skillful technical help. The authors thank David Petroff, Ph.D., biometrician at the clinical trial center of the University of Leipzig, Germany, for proofreading and spell check.

This study was supported by a grant of the German Research Council “Deutsche Forschungsgemeinschaft,” DFG (Bonn, Germany; WR47-1-1). Draeger Medical (Lübeck, Germany) provided an electrical impedance tomography device and the ventilator was supplied by GE Healthcare (Solingen, Germany) without any restrictions, respectively. The University Hospitals of Bonn and Leipzig were supported by research funding from Draeger Medical not related to this study.

Drs. Leonhardt, Putensen, Wrigge, and Reske received lecture honoraria from Draeger Medical (Lübeck, Germany). Dr. Pikkermaat reports financial relationships with Draegerwerk AG and CoKGaA (Lübeck, Germany). Dr. Wrigge reports consultant honoraria from Draeger Medical. The remaining authors declare no competing interests.