Porcine model of sepsis-induced systemic inflammation and acute lung injury in donor lungs


 Background: The shortage of organ donors is a major challenge in lung transplantation. To expand the lung donor pool, ex-vivo lung perfusion (EVLP) has emerged as a platform for assessment and reconditioning marginal donor lungs. In this study a stable and reproducible large animal model of lipopolysaccharide (LPS) induced systemic inflammation and acute lung injury (ALI) was developed.Methods: Pigs (n=6) were anesthetized and monitored. After infusion of LPS (20 μg/kg) for 1 hour, followed by a 90-minute response period, lungs were procured and kept on ice for 2 hours, followed by 4 hours of EVLP. Pulmonary function, inflammatory biomarkers and edema formation were measured in vivo before procurement and during EVLP. Pro and anti-inflammatory cytokines were assayed in blood and in EVLP perfusate, which were collected before and every 30 minutes after LPS administration and EVLP.Results: LPS infusion resulted in significant hemodynamic instability, characterized by marked pulmonary hypertension, decreased systemic blood pressure and increased heart rate. This was associated with increased levels of TNFα, IL-10, IL-6, but no change in IL-1β. Ex vivo assessment of injured lungs showed graft dysfunction characterized by impaired gas exchange and edema formation. The inflammatory profile showed stable but elevated TNFα levels, and continuous production of interleukins during EVLP.Conclusion: We describe a reproducible large animal model of LPS-induced systemic inflammation and ALI. EVLP alone was unable to recondition severely injured lungs. These findings suggest that the EVLP platform requires adjuncts such as targeted anti-inflammatory agents to allow reconditioning of marginal donor lungs.


Introduction
Lung transplantation is an effective therapy for patients with end-stage lung disease 1 . Donor lung availability remains a limiting factor. Based on data from a Canadian multicenter study, only 28% of donor lungs offered were utilized 2 . The most common reason for organ refusal was organ function (79% of lungs offered). However, 27% of those lungs were deemed mildly marginal. It is estimated that two-thirds of these lungs may have been utilized, with or without alternative management increased over 30-40 minutes to 50% CO. When the perfusate temperature reaches 32ºC, ventilation using 50% FiO 2 is started in volume-controlled mode at an initial tidal volume of 4 ml/kg of donor weight, a positive end-expiratory pressure of 5 cmH 2 O and a rate of 5 breaths/min. Tidal volume is increased gradually to a maximum of 6 mL/kg. When a temperature of 37ºC and a target 50% flow are reached, a recruitment maneuver is performed with an inspiratory hold on 20 cmH 2 O over 1 minute and repeated hourly.
At the end of the reconditioning phase, the evaluation phase starts with disconnecting the oxygen supply to the oxygenator, which serves to deoxygenate the perfusate using a gas mixture of 7% CO2 and 93% nitrogen. Lung function is evaluated after ventilation with 100% FiO 2 . Blood gas analyses are performed on samples collected from the left atrium. The pO2/FiO2 (P/F) ratio is calculated.
Electrolytes and metabolites are also measured in these samples and the perfusates collected at the start (T0), at 2 hours and 3 hours of EVLP. The lungs are then weighed (post-EVLP lung weight) to estimate total lung water accumulated during EVLP, followed by a third biopsy (post-EVLP biopsy).

Acute Lung Injury Profiling
The parameters used to evaluate the acute lung injury following LPS infusion include the weight of the graft, wet-to-dry weight ratios; the presence of edema on bronchoscopic evaluation; BALF analysis for total protein. These parameters are measured before and after EVLP.

Inflammation Profiling during EVLP
Kinetics of the cytokine response is explored during two different phases: after LPS infusion in the donor pig and during EVLP. Cytokines are measured in the serially collected arterial plasma samples and perfusates. BALF samples that were collected using bronchoscopy before and after EVLP were also centrifuged and used for the analysis of cytokines. Porcine TNFα, IL-6, IL-10, and IL-1β are analyzed using kits from DuoSet® ELISA Development System (R&D Systems Inc., MN, USA).

Lung Histology
Lung tissue biopsies collected in-situ, at the end of 2-hour cold ischemic time and the end of 4 hour-EVLP periods are fixed in neutral buffered 10% formalin solution for 24-48 hours. Biopsies were embedded in paraffin, sectioned at 5-µm thickness, stained with hematoxylin and eosin, and examined under light microscopy for pathologic changes.

Statistical Analysis
Data in the graphs are presented as means ± standard deviation (SD). The lung weight of the donors in control and LPS groups was compared using the unpaired t-test. Within the LPS donor group, repeated measures during the observation period were compared to baseline data using the paired ttest. For the data collected during EVLP, a 2-way ANOVA test with repeated measures was used to compare the two groups over time. Statistical analyses were performed using GraphPad Prism software (GraphPad Software Inc, La Jolla, CA, USA), and a p-value <0.05 was considered significant.

Hemodynamic changes:
No significant differences were observed between the LPS group and the control group concerning baseline pig weight and ventilation parameters before the LPS infusion. LPS infusion caused hemodynamic instability with an abrupt increase in PAP during the first 10 minutes (supplementary figure 1). This was followed by an increase in systolic blood pressure at 10-15 minutes and heart rate at 20-30 minutes. The baseline mean PAP was 24 ±1 mmHg. The maximum mean PAP (52 ±2 mm Hg) was reached at a mean of 25 minutes and then started to decrease slowly and stabilize at 32±1 mmHg by the end of 1 hour of LPS infusion until lung procurement. Systolic blood pressure and heart rate followed the same trend but lagged by 10 to 20 minutes. The hemodynamic changes were accompanied by a sustained decrease in lung oxygenation capacity by the end of LPS infusion and at started with a higher weight (603 ± 34 g; p<0.001 compared to control lungs) and continued to gain weight (899 ± 76 g; p<0.01 compared to pre-EVLP lung weight) (Figure 6).
The control lungs had increased wet weight during cold ischemia as measured by the wet/dry weight ratio (5.43 ± 0.13 g/g in situ versus 5.70 ± 0.18 g/g pre-EVLP), but reconditioning caused a significant decrease by the end of EVLP (5.17 ± 0.28 g/g versus 5.70 ± 0.18 g/g; p<0.05) (Figure 6). LPS-lungs had higher in-situ wet/dry weight ratio (5.94 ± 0.09 g/g versus 5.43 ± 0.13 g/g in control lungs; p<0.05). In the LPS lungs, no significant change in the wet/dry ratio was observed after EVLP.
Both control and LPS lungs showed minimal edema on bronchoscopy before the start of EVLP reperfusion. By the end of EVLP, bronchoscopy revealed massive edema and flooding in the airways in the LPS group, but not in the control group (supplementary figure 6).

Discussion
This study aimed to develop a reproducible large animal model of LPS-induced systemic inflammation and ALI that could mimic the pathogenesis of human sepsis. This model is different than previously described models because it utilizes a cellular perfusate and an open atrial technique. Previous publications showed similar results using cellular and acellular perfusates in humans 13,14 . We sought to validate the use of cellular perfusate with an open atrial system in a large animal model of sepsis induced lung injury. The protocol is divided into two phases; an in-vivo phase where sepsis is induced with an LPS infusion and an ex-vivo phase, where the lung is procured and placed on EVLP. The present study shows that a model using LPS-induced sepsis yielded an applicable and reproducible platform that can be used to study the effect of sepsis on donor lungs.
During the first phase (in-vivo phase), we assessed the model's ability to reproduce sepsis by examining physiologic, histologic, and molecular parameters. We showed that the model reproduces changes associated with sepsis in all three parameters. Physiologic changes are dominated by the development of pulmonary hypertension in the LPS lungs, which was not present in the control group.
Histologically, we showed that LPS lungs had increased tissue edema and increased weight during the in-vivo phase. Finally, we showed that overproduction of cytokines characterized systemic inflammation in plasma after LPS infusion, mostly TNFα, IL-10, and IL-6. These effects are similar to a more prolonged sepsis model reported by Kubiak et al 15 . Similarly, the same cytokine profile was published in severe human sepsis and acute respiratory distress syndrome in adults by Park 16 . The absence of IL-1β in plasma of pigs after LPS infusion could be expected as previous studies showed no change in IL-1β in humans with severe sepsis 17 .
In the second phase of the study, the ex-vivo phase, we evaluated the lungs using EVLP. We examined donor lungs using physiologic, histologic, and molecular parameters. From a physiologic standpoint, the most striking feature was the increase in PVR that was demonstrated in LPS donor lungs and limited optimal perfusion. In contrast, control lungs had much lower PVR, and optimal perfusion was uniformly reached. In many LPS lungs, target flows were not achieved. The increase in PVR during the ex-vivo phase can be explained by sustained vasoconstriction caused by prolonged action of endothelin, thromboxane and platelet activating factor 18,19 . The net effect of this generalized vasoconstriction in the lungs in the LPS group culminated in hypoxia of cells and deterioration of lung metabolism.
On a molecular level, we observed that cytokines might play an important role in the initiation of lung injury following reperfusion. We noted that during the introduction of perfusion, the cytokine profile is dominated by TNFα, which showed high concentrations in the perfusate and BALF samples. TNFα has been shown to destabilize tight junction proteins and affect cell membrane integrity in pulmonary epithelia, and this may result in pulmonary edema later during reperfusion 20 . Other measured cytokines were also increased by the end of perfusion. In line with these observations, DEVELOP-UK investigators have very recently shown during clinical EVLP that IL-β could be linked to ALI following ischemia-reperfusion 21 . They demonstrated that higher concentrations of IL-1β and TNF-α in perfusate after 30 minutes of perfusion differentiated declined lungs after EVLP from survival lungs.
It should be noted that assessing the effect of EVLP on donor lungs was not a primary aim of our study. Since we do not have a control group that did not undergo EVLP, we can not say whether EVLP does have a therapeutic role in the lungs subjected to sepsis. Instead, we aimed to assess the protocol's ability to reproduce sepsis in a large animal model using EVLP per the Lund protocol. The group from the University of Virginia studied the ability of EVLP to rehabilitate sepsis-induced lung injury 9 . In their protocol, they induced sepsis with an LPS infusion and then randomized pigs to EVLP versus standard procurement and preservation without EVLP. They showed that physiologic and molecular parameters did improve with EVLP. In contrast, we did not note improvements in physiologic or molecular lung parameters with EVLP. Many possibilities may explain the discrepancy in our findings. First, we used a fixed dose of LPS, whereas they used a physiologic determinant to determine the dose given. This may have resulted in a difference in the severity of sepsis in the two studies. Second, we used a cellular prime that contains blood from the donor, whereas they used an acellular priming fluid. Perhaps a cellular prime may further propagate the inflammatory response when compared to an acellular prime. However, previous studies did not show the superiority of acellular perfusate over cellular perfusate 13,14 . Furthermore, a porcine model of extended EVLP up to 24 hours showed an advantage of cellular over acellular perfusates 23 . Both studies were performed in a porcine model with no sepsis. The situation may be different in sepsis-induced lung injury, and perhaps an acellular perfusate may be preferred in this setting. Future applications of this protocol in our laboratory will aim at assessing whether EVLP can be used as a platform to help recondition these lungs, as well as a means to deliver other therapeutic interventions, including ventilatory strategies, molecular targets or other interventions in an attempt to rehabilitate these lungs.

Conclusions
In summary, we believe that our described model can represent a reliable method of inducing and studying sepsis induced lung injury in donor lungs in an animal model. The model includes analysis of donor lungs on a physiologic, histologic, and molecular level. In the future, this model can be used to study the role of EVLP as a therapeutic tool or as a platform for targeted drug and cellular therapy for the amelioration of lung injury in donor lungs.