A complex, reproducible, clinically relevant, in vivo experimental model of severe ARDS induced by two subsequent bronchoscopic instillations of low pH GJ was established in mechanically ventilated pigs. The criteria for severe ARDS was satisfied in all animals, that is PaO2/FiO2 ratio <100 mmHg (Figure 2), bilateral opacities on CXR (Figure E1), pulmonary arterial wedge pressure ≤18 mmHg (Table E1) and lastly the acute onset of injury following GJ instillation present in this study. Despite the severity of the lung injury, the support provided by VV-ECMO allowed the maintenance of adequate gas exchange and stable hemodynamic parameters (Figure 3).
Few other ARDS animal models combined with ECMO support have been described, including injury models with oleic acid infusion, warm saline airway lavage and smoke inhalation[20-22]. However, different from the other published models, our model of severe ARDS is more clinically relevant, as aspiration of gastric contents is a frequent cause of ARDS in clinical practice[1], and causes lung injury with the same mechanisms occurring in the clinical setting. Moreover, ECMO is used as rescue treatment in patients with aspiration ARDS[6], including pregnant women with aspiration pneumonitis after general anesthesia[23]. The consistency and reproducibility of our ARDS model demonstrated by the low variability of the PaO2/FiO2 ratio post injury in the two groups is the result of few technical precautions. The GJ was pulled in one container from different donors, the pH was adjusted at 1.6, and the delivery to the airways was provided through bronchoscopic view in order to specifically target each bronchial segment with a specific volume of fluid.
Our consistent and reproducible model permitted the evaluation of one potential therapeutic strategy that included saline lung lavage combined with SRT early in the course of severe ARDS. Bronchoscopy-based treatments would have otherwise not been possible without extracorporeal support in severely hypoxemic subjects. Indeed, the treatment with saline lavage and SRT was physiologically well tolerated by all the animals in the intervention group (Figure 3), whose gas exchange was maintained in normal range by ECMO. To our knowledge, recent studies on SRT in adult ARDS have not included ECMO patients or included a lung lavage treatment preceding surfactant administration. SRT during ECMO has been studied in pediatric patients and shown to be beneficial[24].
However, the results of our investigation showed that lung physiologic and biologic parameters were not significantly different in treated animals compared with controls (Figures 3).
A large number of clinical studies focused on the potential therapeutic role of SRT in ARDS, but failed to show a significant effect on mortality[10, 11]. Reasons that may explain the negative results include dosing of surfactant, administration modalities, and lastly the persistent presence in the alveolar space of inflammatory factors, which can cause endogenous and exogenous surfactant dysfunction. Moreover, studies on SRT in adult ARDS have not included severely hypoxemic patients, who may benefit the most from any potential ARDS therapy given the severity of lung injury, but for the same reason would not safely tolerate intra-tracheally delivered therapies.
These issues were addressed in our experimental model. Firstly, we caused severe ARDS requiring VV-ECMO to restore adequate gas exchange and stable physiological conditions to tolerate lung lavage with high volume of saline (10ml per bronchial segment, for a total of ~200 ml, with a return of ~100 ml). The ECMO support secondarily allowed comprehensive bronchoscopy in order to remove aspiration contents, inflammatory mediators, and aspiration-induced dysfunctional surfactant, followed by delivery of high doses of exogenous surfactant in each bronchial segment (~5mL (containing 135mg phospholipid)/kg body weight).
A similar approach was studied by Nakajima and colleagues[16] in a lung transplant related experimental model to treat mild acid aspiration-induced lung injury (PaO2/FiO2 ratio 200-300 mmHg) caused in vivo by bronchoscopic instillation of gastric juice. Lungs were treated ex vivo in the EVLP system, which allowed the accurate and safe administration of the therapy independently of gas exchange and the potential associated systemic complications. The results showed that only the combination of lung lavage and SRT, but not lung lavage or SRT alone, resulted in better physiologic lung function and reduced inflammation at the end of EVLP and after lung transplant.
Our study attempted to translate whether this ex vivo approach had broader clinical implications for ARDS treatment, such as in an in vivo setting using VV-ECMO as a platform. Although we employed a similar model of lung injury and a similar therapeutic strategy with lung lavage and SRT, several features in our model may explain the different results from Nakajima and colleagues work. First, the severity of lung injury was considerably higher in our model, as only mild ARDS was achieved in Nakajima and colleagues based on PaO2/FiO2. The more severe lung consolidation in our model may have prevented the exogenous surfactant to adequately reach the alveolar space. Second, the absence of chest wall in the EVLP system may have facilitated lung recruitment with consequent higher exogenous surfactant bioavailability in the alveolar space. Indeed, SRT in combination with lung RM has been shown effective to improve oxygenation and lung volume[24-26]. It would be hence interesting to investigate whether SRT is more effective in ARDS subjects with higher alveolar ‘recruitability’ compared to subjects with persistent lung consolidation. Third, due to the severity and extension of lung injury in our model, the dose of exogenous surfactant may have been insufficient, or the lung lavage may have not been as efficient to remove aspiration contents and the products of the consequent pulmonary inflammatory response. Indeed, in our model the total BA concentrations from BAL, although lower in the treatment group (Figure 6), were not found to be significantly different from controls. Perhaps performing the lavage with surfactant itself, as suggested by the results in a lung contusion model of ARDS[17], could take advantage of its adsorption properties and facilitate distribution and subsequent recovery. However, even exogenous surfactant could have been degraded by the activity of specific enzymes, including the secretory phospholipase A2[27], which in patients with direct forms of ARDS has been shown to inversely correlate with PaO2/FiO2 ratio and mortality[28]. Alternatively, it is possible that lung lavage itself had worsened the injury in the peripheral, ventilated alveolar units, increasing lung consolidation and preventing alveolar delivery of surfactant, or increasing the air-water surface tension, which is recognized as one of the mechanisms of cellular damage and lung injury propagation[29]. By these mechanisms, lung lavage may have also caused an injurious response to the recruitment manoeuvre only in the intervention group. Finally, while in our model lungs were physiologically perfused with blood, which may sustain the inflammatory response to the acute insult in the lung, in the EVLP system lungs are perfused with an acellular solution, which may blunt inflammation and facilitate lung healing.
Our study has a number of limitations. The complexity of the model and the amount of resources required to perform the experiments restricted the number of animals included in each experimental group. A dose response evaluation with different amount of saline for lung lavage and increasing doses of surfactant for the SRT was not performed. Thus, an optimal dose for efficacy in this model was not determined. Although our rationale for our dosage stemmed from computational data by Filoche and colleagues[30], the work of Nakajima et al[16] and the manufacturer recommendation, it may have been inadequate in our experimental model. Our experimental design and timing may have also influenced the observed results. The duration after lavage and SRT that the animal was monitored was relatively short and thus may have precluded the possibility of observing a beneficial effect from the therapy. Our protocol monitored the animal for 5 hr after therapy (4hr on ECMO/1hr off). Previous surfactant studies, instead, monitored subjects for extended periods, often past 4 hr after surfactant was administered[13, 14, 17, 18, 25, 26]. Furthermore, in studies where bronchoscopically administered surfactant did show improvements in oxygenation, benefits were observed >24 hr after treatment[13, 18, 26]. Thus, longer follow up after SRT during ECMO will need to be investigated in future studies.
Further investigations should also address whether different timing and doses of the treatment strategy, including treatment with surfactant replacement only, may be effective in reducing injury and facilitate lung healing. Moreover, the effect of different mechanical ventilation strategies, resulting in better alveolar recruitment, could potentially improve the distribution of the surfactant to the injured areas of the lung. Alternatively, it is possible that the treatment with saline lavage and surfactant replacement is not efficacious in this aspiration model of severe ARDS.