Effects of Ultraprotective Ventilation, Extracorporeal Carbon Dioxide Removal, and Spontaneous Breathing on Lung Morphofunction and Inflammation in Experimental Severe Acute Respiratory Distress Syndrome

PROTECTIVE mechanical ventilation (P-MV) with low tidal volume (V_T, 4 to 8 ml/kg of predicted body weight) and distending pressures (inspiratory plateau pressure) lesser than or equal to 30 cm H₂O reduces mechanical stress to lung tissue, decreasing lung inflammation, and improving survival in patients with the acute respiratory distress syndrome (ARDS).^1,2  However, even low V_T cannot avoid increased lung stress/strain, leading to ventilator-induced lung injury (VILI).³  In fact, in a group of ARDS patients mechanically ventilated with V_T of 6 ml/kg, tidal hyperinflation could still be detected,⁴  suggesting that P-MV in those patients would require even decreased V_T. However, carbon dioxide retention and respiratory acidosis may pose a limit to further reduction of V_T.

Extracorporeal carbon dioxide removal (ECCO₂-R) allows reduction of V_T beyond the threshold of 4 ml/kg (ultralow V_T), while keeping Paco₂ in a clinically acceptable range.^5,6  Nevertheless, reduced alveolar ventilation with ultralow V_T may favor alveolar collapse and further deterioration of oxygenation. Furthermore, most commercially available devices for ECCO₂-R provide adequate carbon dioxide elimination, but not oxygenation.⁷ 

To our knowledge, the role of ultra-P-MV (UP-MV) combined with ECCO₂-R in severe ARDS has not been definitely demonstrated. Furthermore, the effects of spontaneous breathing activity to counteract lung collapse and improve oxygenation during ultralow V_T ventilation have not been investigated.

Spontaneous breathing activity during mechanical ventilation may reverse alveolar collapse, redistribute ventilation and perfusion, and decrease cyclic collapse and reopening of alveoli, possibly leading to less VILI.⁸  On the other hand, spontaneous breathing activity may also result in unpredictable inspiratory effort, increasing stress/strain, and worsening VILI.⁹ 

In this study, we evaluated the effects of UP-MV with and without spontaneous breathing activity on gas exchange, lung mechanics, hemodynamics, regional distribution of ventilation and perfusion, as well as on proinflammatory response, and histological damage in lungs, in a double-hit model of severe early ARDS in pigs. We hypothesized that during UP-MV combined with ECCO₂-R: (1) lung inflammation and damage are reduced compared to P-MV; (2) spontaneous breathing activity, whether supported by pressure or not, enhances oxygenation and further improves lung protection.

After approval by the governmental animal care committee (Landesdirektion Dresden, Dresden, Saxony, Germany), 28 pigs with mean body weight of 41.8 kg (32.8 to 52.5 kg, German landrace) were used for this study.

Animals were premedicated intramuscularly with 10 mg/kg ketamine (Ketamin-ratiopharm; Ratiopharm, Ulm, Germany) and 1 mg/kg midazolam (Midazolam; Ratiopharm), intubated with a cuffed 8.0-mm internal diameter endotracheal tube and mechanically ventilated (EVITA XL; Dräger Medical, Lübeck, Germany). Anesthesia was maintained by means of continuous intravenous infusion of midazolam (1 to 2 mg kg⁻¹ h⁻¹) and ketamine (10 to 20 mg kg⁻¹ h⁻¹). Muscle paralysis was achieved by continuous administration of atracurium (1 to 2 mg kg⁻¹ h⁻¹). Animals were kept in the supine position during the whole experiment. Volume status was maintained with a continuous infusion of Ringer’s acetate (RA—Ringer-Acetat-Lösung Bernburg; Serumwerk Bernburg AG, Bernburg, Germany) at 10 ml kg⁻¹ h⁻¹.

Until induction of ARDS, animals were ventilated in volume-controlled mode with the following settings: fraction of inspired oxygen (Fio₂) = 1.0; V_T = 10 ml/kg; positive end-expiratory pressure (PEEP) = 5 cm H₂O; inspiratory to expiratory time ratio (I:E) = 1:1; the respiratory rate (RR) was adjusted to achieve a Paco₂ in the range of 35 to 45 mmHg.

External jugular vein and internal carotid artery were cannulated with 8.5 French sheaths. The arterial line was used for continuous blood pressure measurements and blood sampling. A pulmonary artery catheter (Opticath; Abbott, Abbott Park, Chicago, IL) was advanced through the venous sheath into the pulmonary artery for continuous measurement of pulmonary arterial blood pressure, mixed venous blood sampling, and cardiac output measurements. The airflow signal was acquired from the internal flow sensor of the ventilator through a serial interface. The airway pressure (P_aw) was measured at the proximal end of the endotracheal tube with a T-piece connected to a differential pressure transducer (163PC01D48-PCB; Sensortechnics GmbH, Puchheim, Germany). Esophageal pressure (P_es) was measured with a balloon catheter (Erich Jaeger, Höchberg, Germany) that was advanced into the mid chest and connected to another differential pressure transducer (163PC01D48-PCB, Sensortechnics GmbH). For acquisition of airway flow, as well as airway and esophageal pressures, a LabVIEW-based data acquisition system (National Instruments, Austin, TX) was used, as described elsewhere.¹⁰ 

Arterial and mixed venous blood samples were analyzed using a standard blood gas analyzer (ABL 505; Radiometer, Copenhagen, Denmark). Oxygen saturation and hemoglobin concentration were measured using an OSM 3 Hemoximeter (Radiometer) calibrated for swine blood. Heart rate, mean arterial blood pressure, central venous pressure, and mean pulmonary arterial pressures were measured using a standard monitor (IntelliVue Patient Monitor MP 50 Philips, Böblingen, Germany). Cardiac output was measured via the pulmonary artery catheter as the average of three repeated injections of 10 ml iced saline into the proximal lumen.

Respiratory signals were acquired at a sample frequency of 200 Hz, using an A/D-card (NI USB-6210; National Instruments) connected to a laptop. Extraction of respiratory variables was performed offline from 45 min recordings of airflow, P_aw, and P_es at each time point. Transpulmonary pressure (P_L) tracings were computed as P_aw minus P_es, whereby peak and mean values were calculated cycle-by-cycle (P_L,mean and P_L,peak, respectively) in all cycles (spontaneous, mixed, and mandatory). During controlled mechanical ventilation, the resistance and elastance of the respiratory system (E_rs and R_rs, respectively) were calculated using the equation of motion, as shown in equation E1:

with airway pressure P_aw, airway flow , volume V, time t, and the total airway pressure at end-expiration P₀.

The distribution of ventilation was assessed using electric impedance tomography (EIT—EIT Evaluation Kit 2; Dräger Medical) as described elsewhere.¹¹  Shortly, a flexible belt equipped with 16 electrodes was mounted at the xiphoid level around the thorax to perform EIT. The output images were recorded at 20 frames/s, during 5 min. Impedance distribution was reconstructed offline using dedicated EIT software (Dräger EIT Data Review; Dräger Medical AG, Germany). Each frame consisted of 32 × 32 image values I (x, y), which were analyzed with a custom-made software as described elsewhere.¹² 

Regional pulmonary blood flow was marked with intravenously administered fluorescent, color-labeled microspheres as described in detail elsewhere.¹³  A different color was administered at Baseline 2 and Time 6 to mark regional perfusion. Postmortem processing of lungs was performed as previously described.^13,14  Briefly, the left lung was flushed, air dried, coated with one-component polyurethane foam (BTI Befestigungstechnik, Ingelfingen, Germany), suspended vertically in a square box, and embedded in rapidly setting urethane foam (polyol and isocyanate; Elastogran, Lemförde, Germany). The foam block was cut into cubes and each cube was weighed and assigned a three-dimensional coordinate. The fluorescent dye was retrieved and read in a luminescence spectrophotometer (LS-50B; Perkin-Elmer, Beaconsfield, United Kingdom). The measured intensity of fluorescence in each probe was normalized according its own weight using equation E2:

Where is the weight-normalized relative pulmonary blood flow of the probe i; x_i is the obtained fluorescence probe i, W_i is the weight of the probe i, and n is the total number of probes. The distribution of pulmonary blood flow along the dorsal–ventral and caudal–cranial axes at each experimental condition was assessed by means of linear regression. Changes in the angular coefficients were used to characterize redistribution of perfusion along the respective axis.

In groups with ultraprotective ventilation, a 15 French and a 17 French catheter (Novalung; Heilbronn, Germany) were placed in the femoral artery and vein, respectively, and connected to an interventional lung assist device (ILA^® Novalung) for ECCO₂-R. A mixture of oxygen and air was used as sweep gas, whereby the gas flow was titrated to Paco₂ = 50 to 70 mmHg. The oxygen fraction of the sweep gas was set to keep the partial pressure of oxygen in the blood flowing across the ILA^® approximately constant, minimizing the membrane oxygenation effect.

Experimental ARDS was induced with a double-hit consisting of saline lung lavage and mechanical ventilation with high V_T. Saline lung lavage (first hit) was performed until Pao₂/Fio₂ was less than 200 mmHg for greater than or equal to 30 min. Following that, VILI (second hit) was performed with the following settings: driving pressure of 60 cm H₂O, PEEP = 0, RR = 10 per min, for 5 min. Lung injury was considered stable, when Pao₂ did not increase within 15 min.

The study was a prospective, randomized multiple arms study, evaluating the effects of four different ventilatory approaches, namely: (1) protective controlled MV according to the ARDS network (P-MV_contr); (2) controlled UP-MV (UP-MV_contr); (3) UP-MV with mandatory cycles and superposed unassisted spontaneous breathing (UP-MV_spont); and (4) continuous positive airway pressure combined with pressure supported (PS) spontaneous breathing (UP-MV_PS).

Figure 1 shows the time course of interventions. After instrumentation, baseline measurements were obtained (baseline 1), and experimental ARDS was induced. Following that, the ventilator settings of baseline 1 were resumed, a stabilization period of 15 min was maintained and measurements were performed (injury). P-MV was initiated in the airway pressure release ventilation mode with the following settings: inspiratory airway plateau pressure (P_aw,plat) targeted at V_T = 6 ml/kg, PEEP = 16 cm H₂O, I:E = 1:1, and RR ≤35 per min to pHa >7.30. V_T was reduced up to 4 ml/kg targeting at P_aw,plat ≤30 cm H₂O. If RR was 35 per min and severe respiratory acidosis with pHa between 7.15 and 7.20 developed, V_T and P_aw,plat were not further reduced. A stabilization period of 30 min was allowed and measurements taken (baseline 2 [BL2]). After BL2, a continuous infusion of heparin at a rate of 25 IU kg⁻¹ h⁻¹ including a loading dose of 80 IU/kg was started. Animals were then randomly assigned to one of the four modes of mechanical ventilation using sealed envelopes. In UP-MV_contr, UP-MV_spont, and UP-MV_PS groups, animals were instrumented and connected to the ILA^® device. In P-MV_contr, a period of sham ventilation of 60 min was maintained to match the time needed for instrumentation and placement of the ILA^® device in the other groups.

Ventilator settings in UP-MV_contr, UP-MV_spont, and UP-MV_PS groups were as follows: airway pressure release ventilation mode with driving pressure titrated to V_T ≈3 ml/kg, PEEP = 16 cm H₂O, I:E ratio titrated to a mean airway pressure (P_aw,mean) equivalent to P-MV_contr, and RR = 15 per min. RR was reduced in order to restraint the mechanical stress inflicted by cycling of the ventilator, that is, the stress rate, which has been shown to influence VILI.¹⁵  In UP-MV_spont and UP-MV_PS, muscle paralysis was stopped and spontaneous breathing resumed. In P-MV_contr and UP-MV_contr, another period of sham ventilation of 30 min was allowed to match the time of resuming spontaneous breathing in the UP-MV_spont and UP-MV_PS groups. In UP-MV_spont, animals were able to breathe spontaneously throughout the whole respiratory cycle. In UP-MV_PS, as soon as signs of spontaneous breathing efforts were observed in P_es tracings, the ventilator was switched to continuous positive airway pressure with PS with following settings: continuous positive airway pressure equivalent to P_aw,mean during P-MV_contr, PS adjusted to V_T ≈3ml/kg. Fio₂ was maintained at 1.0 in all groups throughout the whole experiment. During a period of 6 h, measurements of gas exchange, hemodynamics, respiratory variables, and distribution of ventilation were performed once every hour (Times 1 to 6).

At the end of the observation period, heparin was administered (1000 IU/kg iv) (Ratiopharm) and animals were killed by iv injection of 2 g thiopental (Inresa, Arzneimittel GmbH, Freiburg, Germany) and 50 ml KCl 1 M (Serumwerk; Bernburg, Germany). Lungs were removed under continuous positive airway pressure equal to the PEEP level for further processing. Samples from gravitationally dependent (dorsal) and nondependent (ventral) areas of the right lower lung lobe were snap-frozen in liquid nitrogen and stored at −80°C until further analysis.

For analysis of wet/dry ratio, the right middle lobe was weighted (wet weight) and dried afterward in a microwave as described elsewhere (dry weight).¹⁶  The wet-to-dry ratio was then calculated. Between weighing procedures, broncho-alveolar lavage fluid of the right middle lobe was obtained from three repeated instillations (in-and-out) using 50 ml 0.9% saline solution. The material was centrifuged for 15 min with 200 gauge at 4°C and aliquots of the supernatant were obtained and kept frozen at −80°C until processing.

For histology, the right upper lobe of the lung was perfused with 4% buffered formaldehyde solution although a continuous positive pressure equivalent to the PEEP value during the observation period was maintained at the airway. Lung tissue samples of approximately 8 cm³ were taken from ventral and dorsal zones of the right upper lobe. After perfusion fixation and immersion in 4% buffered formaldehyde solution for 7 days, tissue samples were embedded in paraffin, cut in slices of 5 µm thickness, and stained with hematoxylin–eosin for further analysis. Photomicrographs at magnifications of ×25, ×100, and ×400 were obtained from four nonoverlapping fields of view per section using a light microscope. Diffuse alveolar damage (DAD) was quantified by one of the authors (M.K.), who is an expert anatomist and was blinded to therapy groups, using a weighted scoring system, as described elsewhere.¹⁷  Briefly, values from 0 to 5 were used to represent the severity of seven features of DAD, that is, alveolar edema, interstitial edema, hemorrhage, inflammatory infiltration, epithelial destruction, microatelectasis, and overdistension, with 0 standing for no effect and 5 for maximum severity. Additionally, the extent of each feature characteristic per field of view was determined with values of 0 to 5, with 0 standing for no appearance and 5 for complete involvement. The cumulated DAD Score was calculated as the sum of product of severity and extent of all features, being situated in the range, 0 to 175.

Total RNA from lung was isolated with TRI reagent (Sigma–Aldrich GmbH, Deisenhof, Germany) according to the manufacturer’s protocol, followed by purification with NucleoSpin RNA II columns (Macherey&Nagel, Düren, Germany). The complementary DNA was synthesized with the Revert AidTM H Minus First Strand Synthesis Kit (MBI Fermentas, St. Leon Roth, Germany) from 1 µg total RNA according to instructions of the fabricant. The messenger RNA expression of the inflammatory mediators and markers tumor necrosis factor-α, interleukin 6 and 8 (IL-6 and IL-8), amphiregulin and tenascin-c was quantified using quantitative real-time polymerase chain reaction (Maxima SYBR Green qPCR MasterMix:, Fermentas, St. Leon Roth, Germany) with the iCycler MyiQ2 real-time polymerase chain reaction system (BioRad; Munich, Germany), with cyclophilin A and β2-microglobulin as housekeeping genes. The total protein content in broncho-alveolar lavage fluid and lung tissue was measured using the BioRad Protein Assay (BioRad). Protein levels of tumor necrosis factor-α, IL-6, and IL-8 were measured in lung tissue using commercial ELISA kits (R&D Systems, Wiesbaden, Germany) according to the manufacturer’s instructions. Myeloperoxidase activity in broncho-alveolar lavage fluid was measured using a spectrophotometric assay using 50 mM potassium phosphate (pH 6.0) containing 0.167 mg/ml o-dianisidine dihydrocholrid and 0.0005% hydrogen peroxide.

The sample size calculation for testing the primary hypothesis (UP-MV combined with ECCO₂-R reduces cumulative DAD score) was based on effect estimates obtained from pilot studies. Accordingly, we expected a sample size of seven animals per group to provide the appropriate power (1-β = 0.8) to identify significant (α = 0.05) differences in DAD Score, taking a mean difference of 15 ± 8, two-tailed test and multiple comparisons (n = 6) into account (α* = 0.0083, α* Bonferroni adjusted).

Data are presented as mean ± SD, unless stated otherwise. For functional variables, comparability of groups at injury and BL2 was tested with one-way ANOVA followed by Bonferroni post hoc test. P values were adjusted for multiple comparisons according to Bonferroni. Differences among and within groups (time effect T1 to T6) were tested with general linear model statistics using BL2 as covariate, and adjusted for repeated measurements according to the Sidak procedure. To test DAD Score, we used a linear mixed model for repeated measures (compound symmetry, repeated covariance type), including field of view and region (ventral vs. dorsal zones) as repeated, independent variables, treatment as fixed, independent variable, as well as their significant interactions, to analyze differences in the dependent variable DAD score. Adjustments for repeated measures were performed according to the Tukey Kramer procedure. Residual plots were used to examine model requirements. Other comparisons were explorative in nature. Inflammatory mediators and markers of cell stress were analyzed using Kruskal–Wallis test followed by pairwise Mann–Whitney U test with post hoc adjustment according to Bonferroni–Holm procedure. Statistical analysis was performed using SPSS (v. 17.0, Chicago, IL) and SAS (v. 9.2, procedure mixed, SAS Institute, Cary, NC). Statistical significance was accepted at P value less than 0.05.

Due to technical problems with the EIT device, values were obtained from 24 animals in total (P-MV_contr 7, UP-MV_contr 6, UP-MV_spon 5, and UP-MV_PS 6 animals, respectively). Further measurements were performed in all 28 animals (n = 7 per group). As depicted in table 1, P-MV_contr resulted in average V_T ≈5 ml/kg, and P_aw,peak ≈33 cm H₂O. During UP-MV, V_T and P_aw,peak were further reduced to less than 4 ml/kg and less than 30 cm H₂O, respectively. UP-MV_spont was associated with decreased P_aw,peak compared to UP-MV_contr and UP-MV_PS. P_aw,mean was comparable between P-MV_contr and UP-MV_contr, but higher than UP-MV_spont and UP-MV_PS. During UP-MV_PS, P_aw,peak remained fairly constant, indicating that adjustments of PS were not necessary. P_L,mean did not differ significantly among groups, but P_L,peak was decreased during UP-MV_spont as compared to P-MV_contr. During P-MV_contr, RR and minute ventilation were higher than in other groups. E_rs and R_rs were comparable during P-MV_contr and UP-MV_contr, and pressure–time product did not differ significantly between UP-MV_spont and UP-MV_PS.

The double hit injury resulted in Pao₂/Fio₂ less than 85 mmHg in all animals. As shown in table 2, UP-MV_contr was associated with decreased oxygenation and higher intrapulmonary shunt levels compared to P-MV_contr. The time needed to resume spontaneous breathing was 34 ± 14 and 33 ± 11 min in UP-MV_spon and UP-MV_PS, respectively. Both UP-MV_spont and UP-MV_PS yielded higher Pao₂/Fio₂ and lower intrapulmonary shunt than UP-MV_contr. ECCO₂-R reduced Paco₂ and increased pHa, as compared to P-MV_contr. Heart rate, mean arterial blood pressure, and cardiac output did not differ significantly among groups, whereas mean pulmonary arterial pressures was decreased during P-MV_contr, UP-MV_spont, and UP-MV_PS than UP-MV_contr. Also, central venous pressure was higher during UP-MV strategies. The partial pressure of oxygen gradient across the ILA^® membrane was significantly higher than zero during UP-MV_contr, but not during UP-MV_spont and UP-MV_PS, while the partial pressure of carbon dioxide gradient was always higher than zero in all ultraprotective strategies.

Figure 2 shows the distribution of ventilation. UP-MV_spont and UP-MV_PS were associated with a redistribution of ventilation from central to dorsal lung zones compared to P-MV_contr and UP-MV_contr. However, we could not detect a redistribution of perfusion (differences of angular coefficients of relative pulmonary blood flow between Time 6 and BL2, median [interquartile range]) neither along the ventral–dorsal axis (P-MV_contr: 0.0019 [0.0000, 0.0042]; UP-MV_contr: 0.0007 [−0.0007, 0.0020]; UP-MV_spont: −0.0008 [−0.0022, 0.0006]; UP-MV_PS: −0.0009 [−0.0016, 0.0032]), nor along the cranial–caudal axis (P-MV_contr: −0.0010 [−0.0016, 0.0001]; UP-MV_contr: 0.0006 [−0.0001, 0.0014]; UP-MV_spont: −0.0006 [−0.0015, 0.0019]; UP-MV_PS: −0.0002 [−0.0010, 0.0010]).

As depicted in figure 3, UP-MV_contr reduced the DAD score in dorsal areas, as compared to P-MV_contr, mainly due to decreased alveolar edema and inflammatory infiltrates (table 3). The wet-to-dry ratio did not differ significantly among groups (P-MV_contr: 8.5 [7.8 to 8.9]; UP-MV_contr: 7.8 [7.0 to 9.7]; UP-MV_spont: 7.5 [7.4 to 7.7]; UP-MV_PS: 7.7 [7.1 to 8.0]).

UP-MV_PS was associated with higher levels of tumor necrosis factor-α and IL-8 both in ventral and dorsal lung regions compared to other groups (fig. 4). No differences were found in markers of inflammation in lung tissue among P-MV_contr, UP-MV_contr, and UP-MV_spont. Gene expression of inflammatory mediators and markers of cell stress in lung tissue (table 4), as well as total protein, cytokine, and myeloperoxidase levels in broncho-alveolar lavage fluid (table 5), were comparable among different groups.

In a model of severe ARDS in pigs, we found that: (1) UP-MV_contr reduced DAD score mainly in dorsal lung zones, but worsened oxygenation and intrapulmonary shunt, compared to P-MV_contr; (2) UP-MV_spont and UP-MV_PS improved oxygenation and intrapulmonary shunt, and redistributed ventilation towards dorsal areas, as compared to UP-MV_contr; (3) UP-MV_PS resulted in more inflammation in lung tissue than P-MV_contr, UP-MV_contr, and UP-MV_spont, mainly in dorsal zones.

To our knowledge, this is the first study investigating the impact of different ventilatory strategies, including spontaneous breathing, during UP-MV and ECCO₂-R on lung morphofunction and inflammatory markers in a model of severe ARDS. We used a double-hit consisting of saline lung lavage and VILI, which reproduces most histological features seen in human ARDS.^18,19  The levels of hypoxemia were compatible with severe ARDS according to the Berlin definition.²⁰  We chose pressure-controlled and pressure support ventilation because these modes share similar inspiratory flow patterns. Furthermore, in presence of spontaneous breathing, volume assist-control ventilation may yield breath stacking. In P-MV_contr, settings of V_T, RR, and I:E were based on recommendations of the ARDS network. However, in some animals, inspiratory plateau pressure could not be set lesser than or equal to 30 cm H₂O due to severe respiratory acidosis, but P_L,peak was less than 20 cm H₂O during P-MV_contr, a level that appeared to be safe during the ventilation of pigs without lung injury in a study by Protti et al.,²¹  when sufficient PEEP was used, and is far less than the safety limit of 27 cm H₂O proposed in humans elsewhere.²²  The PEEP level was chosen in agreement with the higher PEEP strategy.²  In fact, a recent meta-analysis showed that higher PEEP levels improve survival in patients with severe ARDS.²³  The Fio₂ was kept at 1.0 to allow direct comparison with UP-MV strategies, where accumulation of nitrogen may impair oxygenation.²⁴  In order to minimize differences in P_aw,mean among groups, we increased the I:E ratio in UP-MV_contr and UP-MV_spont, and the PEEP in UP-MV_PS. Indeed, P_aw,mean may impact on gas exchange, hemodynamics and lung injury,²⁵  affecting the comparability among different MV strategies.

The deterioration of oxygenation and intrapulmonary shunt during UP-MV_contr, compared to P-MV_contr, may be ascribed to alveolar derecruitment due to decreased V_T and P_aw,peak, despite comparable P_aw,mean.²⁶  Spontaneous breathing activity, whether PS or not, improved oxygenation and intrapulmonary shunt, reducing also the mean pulmonary arterial pressure. Previous studies have shown that these effects could be explained by redistribution of perfusion towards better aerated, nondependent lung regions^27,28  or recruitment of collapsed, dependent zones.²⁹  In the current study, we found that UP-MV_spont and UP-MV_PS redistributed ventilation towards dorsal areas, although not affecting regional perfusion. These observations suggest that spontaneous breathing activity induced recruitment in those areas due to higher regional P_L, because P_L,peak and P_L,mean were comparable among UP-MV strategies. The decrease in mean pulmonary arterial pressures during P-MV_contr, as well as UP-MV_spont and UP-MV_PS, compared to UP-MV_contr may be explained by the improved oxygenation. The higher central venous pressure during UP-MV strategies could be attributed to the arterial-venous pressure gradient across the artificial membrane, which may have increased the pressure in the inferior cava vein.

Improved oxygenation in UP-MV_spont and UP-MV_PS compared to UP-MV_contr cannot be attributed to oxygen uptake in the extracorporeal gas exchange device. In fact, we found that the partial pressure of oxygen gradient across the artificial membrane was higher in UP-MV_contr than UP-MV_spont and UP-MV_PS, suggesting that the beneficial effects of spontaneous breathing activity during UP-MV on oxygenation were even underestimated.

The decrease of histological damage in dorsal areas during UP-MV_contr, as compared to P-MV_contr, can be explained by two mechanisms: reduced stress/strain, as indicated by decreased P_aw,peak and P_L,peak, as well as V_T, and decreased stress rates, as suggested by decreased respiratory rate.³⁰  The beneficial effects of UP-MV on histological damage could not be detected when spontaneous breathing activity was resumed, despite comparable values of P_L,peak and P_L,mean. This suggests that inspiratory effort might have partially counteracted lung protection. In fact, UP-MV_PS, but not UP-MV_spont, was associated with an increase in markers of inflammation in lung tissue, mainly in dorsal areas. Since I:E was decreased during UP-MV_PS than UP-MV_spont, it is possible that UP-MV_PS favored the collapse/reopening of most lung units. This finding differs from previous studies from our group showing that pressure support reduces lung inflammation compared to controlled MV.^31,32  This difference could be explained by higher severity of lung injury in the current study as compared to previous ones. In fact, spontaneous breathing has been reported to increase lung injury in an experimental model of severe, but not mild ARDS.⁹  However, we cannot exclude that higher PEEP during UP-MV_PS contributed to increase inflammation in our animals. Furthermore, differences in time-cycled versus flow-cycled assisted breaths may have led to different patterns of distribution of regional stress in lungs. Also, the impact of spontaneous breathing on lung injury may differ between protective and ultraprotective strategies. It is worth noting that spontaneous breathing during UP-MV_spont was not associated with injurious values of P_L,peak, which were decreased than during P-MV_contr. It must be kept in mind, however, that P_L derived from P_aw and P_es does not allow distinguishing the distribution of stress within the lungs on a regional level, where local phenomena, such as pendelluft, which can be caused by spontaneous efforts, may contribute to lung injury.³³  During UP-MV_PS, the relatively low RR and pressure–time product suggests animals were not uncomfortable.

Our results support the hypothesis that UP-MV and ECCO₂-R without spontaneous breathing may improve lung protection in the early phase of severe ARDS, as compared to conventional P-MV. This hypothesis is in line with a prospective cohort study showing that V_T less than 6 ml/kg predicted body weight and extracorporeal decarboxylation improved markers of lung protection.⁵  Furthermore, a recent randomized controlled trial suggested that the use of UP-MV combined with ECCO₂-R has the potential to further reduce ventilator-associated lung injury in severe ARDS.⁶  It is worth noting that, despite potential beneficial effects on oxygenation, a relative worsening of lung damage or inflammation occurred with spontaneous breathing. These results suggest that spontaneous breathing should be used cautiously during UP-MV in the early phase of severe ARDS, even when the patient shows low RR and inspiratory effort.

The current study has several limitations. First, the double-hit model does not reproduce all features of the more complex human severe ARDS. Second, the therapy period was limited to 6 h, and we cannot exclude that results can differ in the long term. Theoretically, complications other than VILI could arise from atelectasis mainly with UP-MV_contr, for example hypoxemia, right heart failure due to an increase in mean pulmonary arterial pressure, pneumonia, and difficult weaning, among others. Third, we used an Fio₂ of 1.0 in all groups. Although comparability among ventilation modes was enhanced, the higher Fio₂ level during a relatively long time period may have led to reabsorption atelectasis, possibly increasing collapse and reopening of mid-dorsal and dorsal lung zones. Fourth, the spontaneous breathing during UP-MV was resumed with time-cycled and flow-cycled modes, and our results cannot be directly extrapolated to other assisted ventilation modes. Fifth, our data were obtained in the early phase of severe ARDS. Thus, different findings are possible when spontaneous breathing is applied later in the course of ARDS. In fact, spontaneous breathing during extracorporeal lung support has been successfully used in the late phase of severe ARDS³⁴  and other forms of lung disease.³⁵  Sixth, to avoid derecruitment and maintain comparability among groups, PEEP was kept constant during the therapy period in all groups, contributing to values of P_aw,plat greater than 30 cm H₂O in some animals.

In the current model of severe ARDS in pigs, UP-MV with ECCO₂-R and without spontaneous breathing slightly reduced histologic lung damage, but not inflammation, as compared to P-MV with low V_T. During UP-MV, spontaneous breathing improved gas exchange and distribution of ventilation, but pressure support increased lung inflammation.

This study was supported by grant no. GA 1256/6-1 from the German Research Council (Deutsche Forschungsgemeinschaft, DFG), Bonn, Germany.

The authors declare no competing interests.