Pathophysiology and pharmacotherapeutic efficacies of ARDS requiring critical care in neonatal and post-neonatal infancy are unclear. In ventilated 7-day-old rabbits, direct (intratracheal) LPS-induced ALI was associated with severe pneumonia but moderate death risks whereas indirect (intravenous) LPS-ALI was with moderate lung injury but high deaths due to sepsis associated multiorgan impairment. Surfactant and inhaled NO failed to improve the survival, associated with highly provoked mRNA expression of proinflammatory mediators as multiorgan involved septic lung damage pattern. The pathophysiological and pharmacotherapeutic response patterns of LPS-route dependent worse outcome implicates causal relations underlying the mortality of experimental pediatric ARDS.

Acute respiratory distress syndrome (ARDS) as a severe and persistent hypoxemic respiratory failure (PHRF) is characterized by pulmonary infection, inflammation, high vascular-to-alveolar permeability and ventilation/perfusion mismatching. Its initial stage is acute lung injury (ALI) or mild-to-moderate ARDS. Current knowledge of etiology, pathogenesis and pathophysiology of ALI and PHRF is focused on direct and indirect assaults to the lungs, also known as pulmonary and extrapulmonary causes. These involve bacterial and viral pneumonia, lung abscess, contusion, airway obliteration, sepsis, acute pancreatitis, acid aspiration of gastric regurgitation, or embolism of bone fracture, etc.; and hemodynamic derangement as sepsis and septic shock with multiorgan system dysfunction/failure [1–4]. Its clinical definition has been modified several times, and response to oxygenation, high positive end-expiratory pressure (PEEP) and restricted tidal volume (V_T) in mechanical ventilation (MV) are considered protective as mainstay [5].

There are differences in pathogenesis, pathophysiology, respiratory mechanics, histological and biochemical alterations for triggers attributable to pulmonary (ARDS_p) and extrapulmonary ARDS (ARDS_exp) [6, 7]. In ARDS_p, the primary attack is due mainly to pathogen-associated response pattern in alveolar epithelial and capillary endothelial cells, causing wide spread activation of alveolar macrophages and the inflammatory response, confined mainly in the alveoli and small airway compartment [8, 9]. The damage to alveolar epithelial cells leads to decreased clearance of edema fluid from the alveolar space of injured blood-gas barrier, inactivation or depressed production of pulmonary surfactant (PS) by type II alveolar epithelial cells (ATII), with degenerative, suppurative and necrotic bronchiolitis and alveolitis, and ultimately remodeling with cystic pulmonary fibrosis. In contrast, ARDS_exp is caused by the release of inflammatory mediators from extrapulmonary assaults and organ tissue lesions, which primarily affect pulmonary capillary endothelial cells, then to ATs. The triggers cause massive production of inflammatory cells and mediators into the bloodstream leading to inflammatory cascade and aggregation of inflammatory cells, resulting in an increased permeability in blood-gas barrier, intrapulmonary edema, and microvascular tone dysregulation [8, 9]. Regarding the lung mechanics in ARDS_p, dynamic compliance of respiratory system (Cdyn) is generally reduced even if high PEEP is applied whereas in ARDS_exp Cdyn would be more associated with the management of high permeability of blood-gas barrier and systemic hemodynamic derangement to assist in lung protective ventilation [10].

Postnatal maturation of lung structure and immune system likely influence the age-related response differences in lung and systemic infection susceptibility. This pertains to mechanisms of pulmonary inflammation, injury, reparation and remodeling in pediatric and infant patients to bacterial, viral and other pathogen induced ALI and PHRF, especially in late neonatal and postneonatal infancy [4, 11]. Moreover, it is unclear how the underlying mechanisms of pharmacotherapeutic action would be, with regard to ancillary respiratory medications in critical care for pediatric and neonatal patients at high death risk of ARDS. The consensus agreement on the definition of pediatric ARDS (pARDS), by the Pediatric Acute Lung Injury Consensus Conference (PALICC) clarified incidence, epidemiology, management, prognosis and outcome assessment [4, 5]. Surfactant therapy is recognized as an optimal treatment in lung protective ventilation strategies for neonatal respiratory distress syndrome (RDS) of prematurity [12]. However, in those from late neonatal to early post-neonatal infancy, there is insufficient evidence supporting the therapeutic efficacy of surfactant treatment for ALI and PHRF, even contradictory in interpretation of its efficacy [13]. Similarly, the existing evidence on the effectiveness of inhaled nitric oxide (iNO) in treatment of pARDS is also in debate due to transient improvement in oxygenation without reducing mortality but raised concerns on the extrapulmonary organ system safety [14]. Therefore, we speculate that patients diagnosed as pARDS may differ from adult ARDS, especially in the lung impairment and treatment response patterns with sensitive pathobiological biomarkers as phenotypes to predict outcome and prognosis [11, 14, 15].

In this study, we aimed to establish a high dose lipopolysaccharides (LPS)-induced ALI model, in rabbits at post-neonatal infancy, with two different administration routes, of etiological distinction, by intratracheal or intravenous injection of LPS with MV, to simulate critical care for neonatal and pediatric patients in early infancy. We first hypothesized that severity of ALI should be associated with LPS route and dosage to lead a survival rate or median survival time (ST₅₀) in reasonable length (within 10 h) of V_T-restricted MV [16]. This should enable characterizing the severity of septic ALI as in pARDS, with survival status as a key outcome linking the two different etiologies and associated damage response patterns. We then hypothesized that a combined PS and iNO regimen would exert better effects compared to MV alone [16], and explored their mechanisms of pharmacotherapeutic action. With unexpected outcome thereof, we managed to explain mechanisms of ineffectiveness/failure similar in the pARDS trials [17–19].

In the initial experiment (ITL and IVL) (Supplementary Figures 2, 3; Supplementary Tables 2, 3), we found that the LPS induced ALI severity was not dose-dependent, neither were the survival, phospholipid content in BALF, or the mRNA expression of pro-inflammatory cytokines and mediators in lung tissue. Therefore, to facilitate data analysis and presentation, the groups of different doses of LPS were combined, hence the number of animals, and designated as 20 mg/kg of LPS in average in the ITL, ITLSN, IVL and IVLSN groups, respectively, in the main results for easy comprehension (Figures 1b,c). Average BW was around 115 g, which was approximately doubled from that at term birth (31 days of gestation, with birth weight of 55–65 g in average), and male-to-female was 4:6. Values of W/D were significantly lower in the IVL than in the ITL (Table 2).

Despite decades of efforts to bridge the knowledge gap, effective respiratory therapies with MV and ancillary medications for rescue purpose in pARDS remain to be limited, with high mortality risks to be solved [28]. The strength of current LPS-ALI model in 7-day old rabbits at their early post-neonatal infancy is characterized as simulating the pARDS_p and pARDS_exp by LPS-route associated different lung damage patterns, as shown in the survival length, LIS and pulmonary and extrapulmonary organ impairment. In neonatal and pediatric critical care, as we chose this animal age in maturation, it often faces with severe pneumonia and sepsis as underlying morbidities, with lung functional failure in ALI, PHRF and septic shock as the most fatal situation. The pharmacotherapeutic response patterns observed herewith may not readily explain the complex pathophysiology of pARDS in humans. However, the outcome was verified by Cdyn, survival rate and ST₅₀ (for survival length), LIS, and the surfactant phospholipid metabolic pools. The key variables as biomarkers strengthened understanding of the underlying morbidities and applicability of pharmacotherapeutic action by surfactant and iNO treatment. It is therefore of causal implication as highly clinical problems oriented, in quest of rescue solution for prediction of pARDS outcome, despite there was an ineffectiveness/failure with the combined surfactant and iNO treatment [17].

From the lung damage patterns shown in the two phases of experiments, the direct ALI was associated with severe pneumonia and higher LIS, but moderate death risks; whereas the indirect ALI was with moderate LIS but high death risks due to sepsis associated multiorgan impairment in the lung, liver and kidney. The combined surfactant and iNO failed to improve the survival, associated with highly provoked mRNA expression of proinflammatory mediators as multiorgan involved septic lung damage patterns, especially in the IVLSN. These suggest imbalanced efficacies of the two combined respiratory ancillary medications when the lung impairment severity, hypotension and septic shock compromised ventilation/perfusion matching as underlying pathophysiologic, causative risks.

In addition, improper and insufficient surfactant dosage, surface activity, composition of phospholipids and hydrophobic proteins, SP-B and SP-C, its anti-inhibition potential and metabolic characteristics, partially associated with manufacturing process, may also determine as covariate towards the worse outcome. These issues should be explored to enrich our knowledge based on the current concept and recommendation for pARDS [29, 30]. This LPS-ALI model should be eligible for investigation of any new invention of surfactant preparation in novel components, manufacturing process improvement, or treatment response which would be associated with timing and dosing (loading and repeat doses) and delivery techniques (bolus, diluted lavage, or both), and interaction with other medications in critical care, in combination with pathobiological variables as phenotypic and sub-phenotypic indicators [31].

We found those of the IVLSN group had the shortest average survival time but with higher TPL and DSPC total pools (BALF + lung tissue) (Tables 4, 5). These findings may also be corroborated by sub-group data analysis for that of the 10-h survivors (Tables 4, 5). It is thereon estimated, approximately 63% and 36% of full dose surfactant TPL in the IVLSN and ITLSN groups, respectively, remaining in the whole lung metabolic pool at the end of MV, over the ensuing 7 h MV, or shorter for those of early death, and that the TPL being supposedly catabolized at a rate of 5%/h or 9%/h in average in the IVLSN or ITLSN lungs, respectively. Thus, these findings may be regarded as baseline evidences concerning the surfactant phospholipid metabolism in ALI for future investigation with stable isotope labeled phospholipids to estimate pharmacokinetic and pharmacodynamic efficiency as well. The presence of supposedly exogenous surfactant phospholipid membrane layers with no overt TM in adjacent alveolar space were in contrast to that of the controls (Figure 5). Given that the lung injury patterns (Figures 2–4), and phospholipid pools in those of early deaths and 10-h survivors (Tables 4, 5), the exogenous surfactant should be distributed in the injured lung alveoli, and taken up by ATII in reference to the control group (ITC, IVC). Whether it follows the re-synthesis and secretion by ATII, as reutilization, in the severely injured lungs is questionable and needs deliberate design of experiment to verify.

The histopathological features of lung impairment in ITL and ITLSN groups attained wide spread purulent and necrotic alveolitis and bronchiolitis, in contrast to mild and moderate inflammation in those of the IVL and IVLSN. However, there were disparities in the death risks, with mild-to-moderate deterioration in the lung mechanics measured by Cdyn in both ITL and IVL, but improved only in the IVLSN group but with higher early deaths. These discrepancies may account for the surfactant and iNO treatment ineffectiveness/failure, and should be of clinical implication in using MV-centered approaches to predict risks in immediate progression [12]. Another potentially eligible approach is to use surfactant lavage and subsequently with the bolus surfactant regimen, as we did in a meconium-induced lung injury model of near-term newborn rabbit model with good efficacy [16]. Repeated dosing is a third option, however, it would incur a decrease of already impaired functional residual lung volume, simply due to the presence of lung edema and consolidation. This was evidenced by higher LIS for alveolar injury and inflammation, and TP in BALF of both ITLSN and IVLSN groups, though their DSPC/TP levels were also elevated. As these animals had metabolic acidosis with low pH and very high lactate in the blood at the end of MV, their early death might be related to impaired hypoxemia-associated hemodynamic condition when intravenous volume expander and vasopressor were not applicable due to the small animal body size restriction.

This LPS-ALI model rendered a suitable approach in search of relevant pharmacotherapeutic products alleviating the severity of pARDS in neonates and infants. The treatment response pattern by the combined surfactant and iNO treatment, to the two different routes of LPS administration, was not expected in the original design. However, the resulting therapeutic response patterns with both physiological and pathobiological variables, underscored the disparities of outcome. Notably, by sub-group analysis, it revealed that those in the ITLSN and IVLSN with short survival length had nearly normal PCO₂ but lower pH and very high lactate in blood at the end of MV (Table 3). Nevertheless, those in the ITLSN, IVL and IVLSN subgroups with short survival length had variable total surfactant phospholipid pools (Table 5).

To our surprise, under the high dose of LPS exposure, there were limited elevation in the mRNA expression of pro-inflammatory cytokines and mediators. In general, being a hallmark feature of septic ALI (ARDS_exp), alveolar macrophages, as an innate immune cell type, are engaged in the maintenance of pulmonary homeostasis of immune response [32]. The classic NF-κB pathway activated pro-inflammatory cytokines, such as TNF-α, IL-1β, IL-6, IL-8, inducing neutrophil chemotaxis and activation in alveolar space, ultimately leading to the necrosis and purulent damages of alveolar structure, causing ALI and ARDS [33]. The evolving permeability as main pathological change in the course of lung damage may be centered in the pulmonary vascular endothelia, in which IL-1β, −6 and −8 play important roles through a number of vascular endothelial cell associated mediators [34]. This may be the next target to investigate for understanding the injury severity and protective ventilation strategy.

The unexpected high early death due to suspected hemodynamic derangement in the IVLSN was based solely on the evidence of enhanced mRNA expression in the liver and kidney as extrapulmonary organ injury, to link that in the clinical findings [35, 36]. In the study design, there was no measurement of extrapulmonary organ system for multiorgan dysfunction/failure. Therefore, current study was an attempt of reverse translation to explain mechanisms behind the clinical trial limitation, or failure, specifically for surfactant and/or iNO in pARDS [15, 37]. The biomarkers for defining the lung injury and alleviation by intervention, such as Ang-1, Ang-2, Tie-2 for vascular endothelial cell injury, and those for lung fluid clearance, surfactant production and metabolism, integrity and proliferation of ATs may be extended to understanding the pathogenesis and mechanistic roles underlying the pathophysiology and pharmacotherapeutic action [38].

Nitric oxide (NO) plays a pivotal role in modulating both intrapulmonary vascular tone and inflammatory responses. The concrete evidence of iNO in alleviating the lung parenchymal injury in ARDS is not clear yet [39, 40], and its off-label use in very preterm infants at risk of early bronchopulmonary dysplasia (BPD), tended to be safe and assumed to be active in modulation of inflammation and coagulation [41, 42]. Moreover, our own experience in experimental bacterial ALI and that of clinical report, suggest safety and optimal to the effectiveness of iNO with 30–50% oxygen [43] which may mitigate hyperoxic hazard. Like the off-label use in surfactant, the ancillary pharmacotherapeutic strategies in pARDS deserve their systematic investigation to reach consensus agreement. We thereby speculate it be different from the conventions of surfactant therapy in neonatal RDS of prematurity.

The most prominent limitation was the lack of direct evidence regarding the SN failure in both ITLSN and IVLSN groups. We consider the following indirect evidence to account for the SN failure, especially in IVLSN. First, the LIS_total values in IVL and IVLSN groups were lower than that of ITL and ITLSN, coinciding with that of Figure 4 at 2 and 10 h. It implied that the lung infection and inflammation as well as the treatment response patterns were not the same due to LPS route, representing a timely cause-effect from the commence of SN treatment to the end of ventilation between the two group. These were in contrary to the respective survival curves as shown in Figures 2a, 3a. We therefore infer that the survival discrepancy between these two groups, as were for ITLSN vs. ITL or IVLSN vs. IVL, should be impacted by factors other than the lung injury severity, very probably associated with systemic hypotension and septic shock in the treatment of SN. Although we implemented intermittent ECG monitoring, there had no intravenous infusion, vasopressor use, and blood pressure monitoring, due to tiny vasculature of the animals. Second, through examination of qPCR results, marked over-expression of mRNA of ANG-2, IL-1, -6 and 8 in the lungs, liver and kidney in IVLSN, implying multiple organ dysfunction. These phenomena were modest or moderate in ITLSN group. Third, by comparing the ultrastructure images, and the surfactant phospholipid pools in BALF + LH, of injured lungs in Figures 2, 3, the concurrent composite images in Figure 5 mainly represent lung maturation (as day 7 of postnatal life) with lamellar bodies and tubular myelin in alveolar space of unaffected lungs (ITC, IVC). In LPS group, the alveolar structure with inflammatory epithelial and interstitial cells, alveolar macrophages, often with prominent intracellular lysosomes, showing infection in the LPS-injured lungs. In ITLSN and IVLSN group, the large ring-like phospholipids in alveolar space as considered exogenous surfactant, with the ring diameter much exceeding that of the alveolar epithelial cells, were not found in the control animal lungs (ITC, IVC). Moreover, Figures 4, 5 revealed that the alveolar space and bronchiolar lumen in the LPS group were often filled with protein-rich exudate. This protein-and-cell debris was likely to sequester or inactivate endogenous and exogenous surfactant, thereby jeopardize the latter therapeutic action. In our previous preterm and near-term rabbit study, a porcine lung natural surfactant isolated SP-A strengthened biophysical and physiological properties of a porcine lung derived surfactant phospholipid preparation (Curosurf®, with SP-B and C), by reversing inactivation and inhibition from blood plasma proteins (fibrinogen/fibrin, albumin) and meconium [44]. Thus, this study did not have sufficient evidence to explain the mechanisms of exogenous surfactant therapeutic function in counterbalance inactivation/dysfunction, and iNO safety inferential in hemodynamic homeostasis. It remains a crucial and imperative knowledge gap in comprehension of LPS-ALI injury model simulating neonatal and pediatric critical care.

In addition to above strengths and limitations, there are other limitations to address. For the three different LPS dosing groups in the initial experiment, we did not have discerned the LPS dose dependent lung injury by severity and survival status, but found LPS route-dependent effects in survival. Neither did we found any marked benefits of combined PS and iNO in both ITLSN and IVLSN, with IVLSN the worst in survival, though surfactant phospholipid pool and Cdyn seem to be potentially optimal in the 10-h survivors. It was quite different from our previous experiments in adult or young animals at the expectation to find therapeutic efficacy by MV and ancillary medications with PS and iNO [24, 45]. Another weak point is the lack of intravenous fluid maintenance as the fragile vasculature for indwelling intubation. The intraperitoneal injection followed our previous experience with ventilated near-term rabbits that may in part counter balance acidosis and basal metabolism requirement [16, 20, 46]. In future study, these problems may be dealt with deliberate fluid injection and altering of PS dosage, along with other medications, such as intratracheal glucocorticosteroids, or vasopressors, to maintain a longer survival. The route and dose of LPS may still be an important issue for pharmacotherapeutic efficacy in parallel comparison of ARDS_p and ARDS_exp. This may need a framework to standardize the LPS, or other inducing chemicals, with dosing regimen, more measurement of multiorgan functioning variables including sensitive biomarkers to define. Besides, the high LPS doses used in this study may induce a hyper-acute, fulminant lung injury that does not fully recapitulate the subacute phase of human pARDS. Therefore, the model may be best suited for studying early ALI, rather than its chronic or reparative phases.

This study successfully established pulmonary and extrapulmonary ARDS models induced by a high dose range of LPS, in 7-day-old rabbits, explicitly depicting the distinct pathophysiological mechanisms underlying the pharmacotherapeutic action using the combined surfactant and iNO. The impact was with the disparities of LPS route dependent lung injury and survival patterns, underscoring the pathobiological phenotypes of associated pulmonary impairment and inflammatory responses versus the extrapulmonary multiorgan involvement. The model and concept deserve further validation in the overall and specific efficacy assessment for rescue of pARDS_p and pARDS_exp.