Unilateral acute lung injury in pig: a promising animal model

In pigs, a tracheal bronchus branch, named the bronchus trachealis, originates from the trachea proximal to the carina to aerate the upper parts of the right lung (Fig. 2) [15]. To fit the anatomy of the bronchial tree of pigs, we modified a commercial left-sided double-lumen tube for left bronchial intubation in humans (37Fr, Hudson RCI®, Sheridan® Sher-i-bronch®, Teleflex Medical, Morrisville, NC, USA). This adaption was mandatory to avoid blockage of the bronchus trachealis so that ventilation of the right upper lobe was guaranteed. Use of a commercially available left double-lumen tube would have resulted in atelectasis of the right upper lobe. For this, the tracheal cuff was detached from the double lumen tube at the distal end and everted 4 cm proximally. The cuff was fixed and glued (Silastic®, Medical Adhesive Silicon Type A, Dow Corning Corporation, Midland, MI, USA) to the tube in the proximal position (Additional file 1: Supplement Fig. S1). The cuff was then partially inflated with 2 to 3 mL of air to avoid undesirable adherence of the cuff material to the tube in the adjunct areas of the fixation zone. After a drying period of at least 72 h the cuff was tested for air leakage with a cuff pressure of 30 cmH₂O. If no cuff pressure drop was observed within 5 min the cuff was deemed to be airtight, and the double lumen tube could be used in the pig experiments.

All pigs were premedicated with ketamine 5–6 mg kg⁻¹ i.m., midazolam 0.2–0.3 mg kg⁻¹ i.m. and medetomidine 0.05–0.1 mg kg⁻¹ i.m. and placed in the supine position on a heating mat. After cannulation of an ear vein, propofol 2.5–5 mg kg⁻¹ i.v. and 100 µg fentanyl i.v. was administered after which the pig was orotracheally intubated with a commercial singlelumen endotracheal tube (Rüschelit® Super Safety Clear, Teleflex Medical, Co. Westmeath, Ireland) with an ID of 8.0 mm. Pigs were ventilated using a human ventilator (Dräger Primus®, Dräger Medical, Luebeck, Germany) using a volume-controlled mode with tidal volumes of 6–8 mL kg⁻¹ and positive end-expiratory pressure (PEEP) of 5 cmH₂O, resulting in maximum inspiratory pressures (P_max) of ~ 20 cmH₂O, at a respiratory rate of 12–20 breaths per minute. Anesthesia was maintained with continuous infusion of 2% propofol (20 mg kg⁻¹ h⁻¹), fentanyl (5 µg kg⁻¹ h⁻¹) and rocuronium (1 mg kg⁻¹ h⁻¹). A central venous catheter was placed in the external jugular vein under ultrasound guidance, and an arterial line was placed in the left or right common carotid artery. A suprapubic urinary catheter (Zystofix®, B. Braun SE, Melsungen, Germany) was placed for urinary output measurements under ultrasound guidance. After instrumentation and preoxygenation of the pigs, the commercial single lumen tube was replaced by the modified double-lumen tube (DLTm) over a tube exchange catheter (Cook Medical, Limerick, Ireland). The position of the double-lumen tube was bronchoscopically verified (Pentax® FI-13RBS, Pentax Europe GmbH, Hamburg, Germany) in all animals. The position of the tube was adjusted until the tracheal bronchus, the carina and the left and right main stem bronchus were confirmed. After correct positioning of the double-lumen tube, the left and right lung were ventilated separately using two ventilators, one ventilator for ventilation of the right lung and another ventilator for ventilation of the left lung (Additional file 1: Supplement Fig. S2). For both lungs, we used volume-controlled mode with tidal volumes of 3 mL kg⁻¹ for the left lung and 5 mL kg⁻¹ for the right lung. A positive end-expiratory pressure (PEEP) of 5 cmH₂O, resulted in maximum inspiratory pressures (P_max) of ~ 20–25 cmH₂O in the right lung and ~ 15–20 cmH₂O in the left lung, at a respiratory rate of 12–20 breaths per minute. Then, animals underwent a median sternotomy, and bilateral pleural opening was performed to have free access to the right and the left lung. Lung separation was followed by temporary closure of the thorax with clamps and hypoxic preconditioning of the left lung. Hypoxic preconditioning was performed by transient termination of ventilation of the left lung while ventilating the right lung with 100% oxygen for 10 min. Hypoxic preconditioning was performed to account for the Euler-Liljestrand mechanism [16] and, thus, support hemodynamic stability and to minimize shunt fraction during induction of lung injury. After 10 min of hypoxic preconditioning, pigs underwent three cycles of rinsing of the left lung with 0.9% NaCl and Triton® X-100. Animals in the pilot study underwent rinsing in the supine position, whereas animals in the main study were positioned in the left lateral decubitus position before cyclic rinsing. The correct tube position was reconfirmed via fiberoptic bronchoscopy in all animals of the pilot study and the main study.

A pilot study with nine pigs was conducted as a first step to achieve strictly unilateral ALI in pigs weighing 49–64 kg. This included the optimal production of the DLTm. We evaluated the optimal position of the tracheal cuff to avoid atelectasis of the upper parts of the right lung. In three pigs, the lung block was harvested with the DLTm tube in place, and the distance between the everted tracheal cuff and the tracheal bronchus was measured. We found that a proximal shift of the tracheal cuff of 4 cm was sufficient to avoid blockage of the tracheal bronchus.

Despite optimal positioning of the tracheal cuff and optimal blockage of the left main bronchus with the DLTm spillover of the rinsing fluid occurred in 6 out of 9 pigs, thus injuring parts of the right lung. After unsuccessful production of unilateral conditions of lung injury in 6 out of 9 pigs we decided to perform the rinsing procedure in the left lateral decubitus position. This step proved to be essential in producing strictly unilateral ALI.

Six male Austrian Landrace pigs (Sus scrofa domesticus) with an average body weight of 62 kg (range 56 to 76 kg) were used for the experiments. Cyclic rinsing of the left lung was performed in all pigs of the main study in the left lateral decubitus position. Otherwise, the experiments were conducted in a manner comparable to the pilot study.

Lung injury was induced in all animals of the pilot study and the main study by rinsing the left lung with a mixture of 150 mL 0.9% saline and 0.3% Triton® X-100 (Sigma Aldrich, Saint Louis, MO, USA). Rinsing was performed in the supine position in all animals of the pilot study and in the left lateral decubitus position in all animals of the main study. During the purging of the left lung, continuous bronchoscopic control via the tracheal lumen was used to detect any spill-over of the rinsing solution into the right lung in all animals of the pilot study and the main study. These measures were undertaken to guarantee strict unilateral rinsing. To ensure optimal distribution of the rinsing fluid in the peripheral lung tissue, the left lung was ventilated with increasing P_max of up to 40 cmH₂O for 5 min. Thereafter, the remaining rinsing liquid in the left bronchial system was removed via suction. This rinsing procedure was repeated twice; thus, 3 cycles of rinsing were performed in all pigs. A mean arterial pressure > 70 mmHg was set as the target to guarantee hemodynamic stability during the vulnerable rinsing period. Boluses of phenylephrine (Biorphen®, Sintetica, Muenster, Germany) 0.2 mg were administered, if required.

After induction of lung injury, ventilation of the right lung was adjusted to reach the predefined target values for respiratory and metabolic stability, wherein we targeted a PaO₂ > 60 mmHg, a PaCO₂ < 65 mmHg, and SpO₂ > 90%. The left lung was ventilated using lung-protective ventilation with V_T of 3 mL kg⁻¹, PEEP of 8 cmH₂O and Pmax of < 30 cmH₂O. In cases of desaturations or hypercapnia, ventilatory settings were adjusted. Eight out of nine pigs were ventilated for 7 h in the pilot study (one animal died prematurely). Two pigs were ventilated for one hour, and four pigs were ventilated for seven hours in the main study.

BAL was performed in all pigs (n = 9) of the pilot study and three pigs of the main study for measurement of interleukins (IL-6 and IL-8). All of these animals underwent mechanical ventilation for at least 7 h. No BAL was performed in animals with ventilation time ≤ 1 h. One animal died prematurely and was therefore excluded from the analysis of IL-6 and IL-8 in BAL. Bronchoalveolar lavage of the left and right lung was carried out at fixed time intervals (at 1, 3, 5, and 6 h) to measure typical inflammatory parameters, such as interleukin-6 (IL-6) and interleukin-8 (IL-8). BAL in the left and right lung was performed with a saline bronchial flush of 40 mL saline 0.9%, which was sampled again via suction. Samples were stored immediately on ice and were centrifuged at 2000×g at 4 °C for 10 min, and shock frozen at − 80 °C within 3–5 min thereafter. For analysis, samples were thawed and centrifuged at 1000×g for 20 min. Cytokine levels were determined using commercial ELISA kits for IL-6 (MyBioSource, San Diego, CA) and IL-8 (Thermo Fisher Scientific, Waltham, MA).

At the end of the experiment, pigs were sacrificed with an overdose of fentanyl 500 µg i.v. and 30 mL potassium chloride 7.45% i.v.. Thereafter, the lungs were removed en-bloc. Six lung tissue samples were obtained from standardized areas in the left and right lung (Additional file 1: Supplement Fig. S3), thus, 12 tissue samples were collected from each pig. Lung samples were processed and analyzed by an independent and blinded pathologist to test for histological signs of inflammation and destruction. Then, 7.5% buffered paraformaldehyde was used to fix the lung tissue for at least 12 h. Fixed lung tissue was then cut to size, placed in tissue capsules, drained overnight, and embedded in paraffin the next morning. Sections of two micrometers thickness were cut and stained with hematoxylin–eosin or with trichrome staining for better visualization of fibrin (pink) and erythrocytes (orange) using acid fuchsin orange G aniline blue (according to Mallory & Cason). The tissue samples were stored in paraformaldehyde at 4 °C until evaluated and graded according to the Lung Injury Scoring System (LISS) adapted from Matute-Bello et al., and used previously by other authors for numerical evaluation of lung injury [17].

For the main study we defined LISS as main outcome parameter, because histopathological changes are rated crucial in defining ALI in the consensus paper of the American Thoracic Society [17]. A LISS of zero is associated with healthy lung structure and a LISS of one (1.0) is associated with completely destroyed lung structure. The precondition of strictly unilateral ALI is a substantial difference in LISS between the right and the left lung. Therefore, we demanded a mean difference in LISS between the right and left lung of 0.5 and assumed a rather large standard deviation in differences of 0.2. With alpha of 0.05 a power of 0.9 can be achieved with six pigs in paired samples. Thus, a sample size of six pigs was chosen for the main study.

Because of the small sample size Wilcoxon-signed-rank-test was used for comparison of outcome parameters throughout. Descriptive statistics were expressed as mean ± SD, the significance level was set at p ≤ 0.05. IBM SPSS Statistics software (Version 27.0.1.0) was used for data processing.

Macroscopic signs of lung damage of the left lung were present in all pigs in the pilot study (n = 9) as well as in the main study (n = 6). The right lung showed no macroscopically visible signs of damage in all animals of the main study (Fig. 4), whereas six out of nine animals showed macroscopic signs of right sided lung damage in the pilot study group due to spill-over (Additional file 1: Supplement Fig. S4). Strict unilateral left lung damage occurred in all pigs of the main study independent of the elapsed time between ALI induction and harvesting of the lungs. Comparable macroscopic results were achieved in pigs ventilated for one hour and for seven hours after ALI induction.

Histopathological findings were in line with the macroscopic examination of the lungs. Histopathological analysis showed typical signs of ALI [17] in the left lungs whereas the normal histological structure was preserved in the right lungs (Fig. 5). To ensure a reliable grading of the lung damage, the adapted Lung Injury Scoring System (LISS, Additional file 1: Supplement Table S2) was used for evaluation of all histological samples in the main experiment: 12 specimens per pig, 72 specimens in all six pigs in total. The score ranges from 0 (no detectable lung damage) to 1.0 (maximum destruction of lung tissue). The median score of the histological samples of the left lung was 0.72 (IQR 0.62–0.79), and the median score of the histological samples of the right lung was 0.14 (IQR 0.14–0.16), p = 0.031 (Fig. 6).

A severely impaired gas exchange could not be observed in the arterial blood gas analyses, due to the preserved gas exchange in the right lung targeting at an S_pO₂ > 90%. Dynamic lung compliance (C_dyn) was evaluated since it is severely impaired in ALI. Baseline C_dyn was comparable between the left lung and right lung in all animals (C_dyn > 20 mL cmH₂O⁻¹). A drop in C_dyn < 10 mL cmH₂O⁻¹ occurred in the left lung immediately after ALI induction. This significant decrease persisted over the entire course of the experiment (p = 0.031; Fig. 7). A slight decrease in compliance occurred in the right lung over time, probably due to the high ventilatory pressures required to preserve stable metabolic values.

The values for IL-6 in BAL are depicted in Fig. 8a (n = 11). Four consecutive BAL cycles were performed 1 h, 3 h, 5 h and 6 h after ALI induction. Elevated levels of IL-6 were observed in the BAL specimens of the left lung but also in the specimens derived from the right lung. In BAL1, one hour after ALI induction, the median IL-6 level on the left side was 57.4 pg/mL (IQR 47.8–58.4 pg/mL) versus 46.0 pg/mL (IQR 45.4–54.1 pg/mL) on the right side. IL-6 levels in BAL2 averaged 52.7 pg/mL (IQR 48.9–55.7 pg/mL) on the left side compared to 50.1 pg/mL (IQR 49.6–64.2 pg/mL). In BAL3 IL-6 levels were 48.5 pg/mL (IQR 45.4–53.3 pg/mL) in contrast to 47.7 pg/mL (IQR 42.4–49.9 pg/mL) in the BAL of the left lung. BAL4 exhibited median IL-6 levels of 54.3 pg/mL (IQR 50.9–56.4 pg/mL) in the BALs of the right lung compared to median IL-6 levels of 49.4 pg/mL (IQR 43.2–54.8 pg/mL) in the BALs of the left lung. Overall, there was no significant difference in IL-6 levels in BAL between the left and right lungs (p = 0.365).

Furthermore, IL-8 levels were analyzed at comparable time intervals, as shown in Fig. 8b (n = 11). A steady increase in IL-8 occurred over time on both sides. In BAL1, the median IL-8 level in the left lung was 229.7 pg/mL (IQR 200.9–586.7 pg/mL) compared to 206.1 pg/mL (IQR 170.7–437.1 pg/mL) on the right side. The median IL-8 level in BAL2 of the left lung was 351.2 pg/mL (IQR 206.3–1026.0 pg/mL) compared to 694.3 pg/mL (IQR 269.2–1735.3 pg/mL) in BALs derived from the right lungs. The median IL-8 level in BAL3 was 659.3 pg/mL (IQR 253.2–1303.2 pg/mL) on the left side versus a median IL-8 level of 787.3 pg/mL (IQR 345.1–2170.6 pg/mL) in BAL on the right side. In BAL4 of the left lung, the median IL-8 values were 1001.4 pg/mL (IQR 493.0–1326.3 pg/mL) and 1247.7 pg/mL (IQR 477.6–2340.6 pg/mL) in BALs derived from right lungs. In summary, there was no significant side difference in IL-8 levels detected in BALs from the right lungs in comparison to IL-8 levels detected in BALs from the left lungs (p = 0.206).

We examined the 6 pigs with strict unilateral ALI separately (3 pigs from the pilot study, 3 pigs from the main study) and were not able to determine any significant difference as well (Wilcoxon-signed-rank-test for IL-6: p = 0.563 and for IL-8: p = 0.156).

Typical histological evidence of ALI assesses neutrophilic infiltration of the interstitium and the alveolar space, formation of hyaline membranes and edema of the alveolar walls. These histological findings are summarized in the lung injury score initially introduced by Matute-Bello et al. [17] Most investigators use an adapted form of the score [18‐20]. Since the evidence of hemorrhage is regarded as a somewhat relevant feature we included it in the adapted histological score. The left lung showed macroscopically severe destruction, whereas the right lung proved to be macroscopically unchanged in our model. The adapted histological score showed significant evidence of the typical signs of ALI in the left lung and normal values for the right lung. Therefore, histological evaluation proved the unilateral character of ALI in our model.

Since the animals were sacrificed approximately eight hours after induction of ALI, the typical features of the early, exudative phase of ALI were pronounced, whereas fibroblastic remodeling and collagen deposition in the lung associated with the fibrotic phase were not found [21‐23]. This is typical for the majority of animal models [17] since the exudative phase of ALI lasts approximately 7 days after induction of ALI [24].

Generally arterial oxygen tension (P_aO₂) and P_aO₂/F_iO₂ are used to prove functional impairment and determine the severity of ARDS in humans (the Berlin definition for ARDS) [25]. Due to the unilateral character of our model, P_aO₂ and the P_aO₂/F_iO₂ ratio remain in a physiological range, and thus, are unsuitable parameters for the evaluation of functional impairment in our model. Since the right and left lungs were ventilated with two separate ventilators, dynamic compliance (C_dyn) served as a parameter indicating the onset of severe lung injury. C_dyn dropped within 15 min after cyclic rinsing below 50% of baseline in the left lung and remained in the normal range in the right lung. The drop in compliance indicated rapid functional deterioration of the left lung.

The preservation of P_aO₂ and metabolic values in our animals is an advantage over bilateral models because the incidence of premature death before finishing the experiment was extremely low in our study when compared to recent reports of bilateral ALI in pig models [26]. Repeated lung lavage without detergence is also associated with hemodynamic and metabolic stability but does not lead to severe epithelial damage or nor to a substantial reduction in C_dyn [27], making the model unsuitable for studies investigating strategies of protective ventilation. Models known to reproduce epithelial injury seen in human ARDS/ALI, such as the oleic acid model or the endotoxin model, lead to severe structural damage but cannot be used in a unilateral model. Additionally they are associated with severe hemodynamic and metabolic instability [27]. Therefore, the combination of metabolic and hemodynamic stability in combination with severe histologically proven epithelial damage and an immediate and sustained drop in C_dyn, as observed in our unilateral ALI model, seems to be of advantage.

An immediate decrease in C_dyn is also associated with models using tracheal application of gastric juice or hydrochloric acid (HCl) [28]. Lung injury exhibits a patchy pattern in models using HCl or gastric juice. The patchy pattern can be misleading if small regions of the lung are used to study drug effects, etc. In addition, the amount of lavage fluid administered also seems to be linked to the severity and time course of onset of ALI. Meers et al. used 4 mL/kg (150 mL) for ALI induction in a bilateral ALI model. In our model, 3–4 mL/kg saline 0.9% + Triton® X-100 0.3% was instilled solely in the left lung, and the rinsing was repeated twice. This might, in addition to the use of detergent, explain the more rapid drop in C_dyn in our model compared to the model of Meers et al. In contrast, the porcine ALI model of Tiba et al. was initiated with no more than 1 mL/kg gastric juice producing only mild ALI after 12 h [29].

Another hallmark in ALI models is the measurement of cytokines reflecting the degree of inflammation. Meduri et al. showed that high cytokine levels (TNFalpha, IL-1β and IL-8) during the exudative phase of ARDS negatively affect the severity of ARDS and mortality in humans [30]. In the early period of ARDS alveolar macrophages are activated by proinflammatory chemokines. As a result, toxic mediators are released and lead to a further increase in the inflammatory cascade [21]. The systemic proinflammatory cytokines IL-6 and IL-8 are associated with adverse outcomes in ARDS [31]. Therefore, we assessed IL-6 and IL-8 in our study.

We found similar or even higher interleukin levels in BAL compared to other animal models of ALI [26, 28, 32‐34]. Additionally, the amount of released mediators correlates with the severity and onset of ALI. In a novel model of ALI induction with smoke inhalation [26], IL-6 levels increased no earlier than 48 h after smoke inhalation. In contrast, we observed an increase in IL-6 in BAL as early as 1 h after lung lavage. Additionally, the concentrations measured in our model were 3–4 times higher than those observed by Leiphrakpam et al. [26], indicating a massive inflammatory reaction in our unilateral animal model. In line with this, we also found increased IL-8 levels in BAL. IL-8 levels increased further during the course of the experiment. This time course of IL-8 release is in line with other findings [21, 35].

We detected severely elevated IL-6 and IL-8 levels in the healthy and the injured ALI lung suggesting severe systemic inflammation due to development of unilateral ALI in our animals. Elevation of interleukins was observed as well in the histopathological healthy right lung. Pro-inflammatory mediators cause oxidative stress and lead consecutively to severe damage and death of cells. Oxidative stress can also be a side effect of various drugs. Antioxidants are of enormous research interest to protect against or ameliorate oxidative stress [36‐38] caused by inflammation or drug toxicity. Since our model can be regarded as a model of inflammation and/or toxic injury it offers a potent possibility to test for the protective effects of antioxidants of various kinds.

Various environmental compounds, drugs and xenobiotics induce their deleterious side effects thorough increased reactive oxygen species (ROS) production [39]. Mitochondria and lysosomes were shown to be primary targets for toxicity induced increased membrane permeability associated with the release of free radicals [38, 40]. Our animal model could serve for testing of protective effects of many ROS scavenger agents in healthy and already severely injured lung tissue.

All animal models have that try to mimic human ARDS. However, pig animal models have been reported to show better translational potential due to similarities in anatomy and physiology [41]. Lung injury in ARDS has been reported to develop in three pathophysiological phases: the exudative phase, the proliferative phase, and the fibrotic phase [22]. Since our animals were sacrificed within 8 h after induction of ALI, it can be representative only for the exudative phase of the disease.

Our model does not appear to be able to determine any left-to-right difference in IL-6 and IL-8 despite successfully induced unilateral ALI. Although the levels of the interleukins are much higher than physiological normal values, the progression is more or less parallel when comparing the left and right lung. It can be assumed that the massive inflammation in the left lung leads to a systemically relevant inflammatory reaction and as a result, interleukins are also released in epithelial lining fluid of the right, healthy lung while its function and gas exchange are preserved.

Another potential downside of our unilateral pig model is the technical complexity of inducing unilateral ALI. The adaptions of the commercially available double-lumen tube for humans are not very demanding. However, the bronchoscopic placement of the tube requires an experienced physician because it is key in producing unilateral ALI and avoiding spill-over of lavage fluid to the healthy lung. Additionally, the changes in position (lateral decubitus to supine) require close attention to hemodynamic stability and expert use of vasopressors (norepinephrine and phenylephrine). Furthermore, the experimental animals should weigh between 45 and 65 kg to achieve a correct and stable tube position. Therefore this protocol is unsuitable for piglets.