Integrating molecular pathogenesis and clinical translation in sepsis-induced acute respiratory distress syndrome (original) (raw)

Our current understanding of the biology underlying sepsis and ARDS is vast, and many detailed reviews have recently been published (30, 40–42). We focus on selected findings and contextualize them as they pertain to sepsis-induced ARDS.

Generation of inflammation. Sepsis-induced ARDS is initiated by an inflammatory host response to a microbial pathogen. The host response can be categorized into the innate immune response, which functions via pattern recognition receptors (PRRs) directed against microbial molecules and the adaptive immune response (humoral immunity from antibodies produced by B lymphocytes and cell-mediated immunity mediated by T lymphocytes), as well as host-derived molecules, leading to activation of cellular immune responses.

PRRs are membrane-bound or cytoplasmic proteins that bind to corresponding PAMPs, such as microbial nucleic acids, bacterial cell surface lipoproteins, and LPS. PRRs are present on surveillance cells, such as immune, epithelial, and endothelial cells, that sample the local environment (41). For example, Gram-negative bacteria–derived LPS is a PAMP that binds the cell-surface PRR TLR4, leading to innate immune response activation (43). The PRR family also includes transmembrane proteins, such as other TLRs and C-type lectin receptors, and cytoplasmic proteins, such as NOD-like receptors and retinoic acid–inducible gene-I–like receptors (44). In addition to microbial signals, PRRs recognize endogenous DAMPs released from intracellular and extracellular compartments during sepsis. DAMPs include those of nuclear origin (e.g., DNA, high-mobility group box 1 protein) and those from the cytoplasm (e.g., RNA), the ECM (e.g., hyaluronic acid), or other sites (e.g., mitochondrial DNA [mtDNA], ATP) (42, 45–47). An initial septic insult produces downstream inflammation from PAMP stimulation of the innate immune system. This inflammation is subsequently amplified by DAMP signaling, resulting in further cell and tissue injury (41). These pathways trigger a cycle of injury, leading to alveolar-capillary barrier dysfunction, facilitation of additional pathogen invasion, and PAMP stimulation of the innate immune system.

Early control of infection provides the best defense against the release of PAMPs and DAMPs. Data from clinical studies support early antibiotics as a critical intervention for improving mortality from sepsis and development of end-organ dysfunction. Mortality from sepsis is 5 times higher when the initial antibiotic choice is inappropriate (48), and increased time to antibiotic administration is associated with progression to septic shock (49). Activation of the innate immune response following infection is regulated by various cellular pathways, including autophagy, a highly conserved pathway involved in clearing damaged proteins and organelles from cells. Following activation of the innate immune response, autophagy plays a key role in limiting mtDNA release, which is dependent upon the NALP3 inflammasome (50). Inflammasomes are multiprotein complexes that mediate caspase-1 activation, which promotes maturation and secretion of IL-1β and IL-18. Interestingly, mice lacking key autophagy proteins produced higher levels of IL-1β and IL-18 in sepsis models and had higher LPS- and polymicrobial sepsis–induced mortality than WT controls (51–53).

Our group and others have observed a striking association between elevated blood mtDNA levels and mortality in sepsis and ARDS, as well as an association between IL-18 levels and mortality (54–56). These data suggest that targeting inflammasome and autophagy pathways may represent a novel therapeutic strategy. LPS (via TLR4 stimulation) can also induce autophagy through a p38/MAPK/VPS34 pathway that results in activation of heme oxygenase-1 (HO-1), a cytoprotective molecule upregulated in sepsis and ARDS (57). CO, a gaseous molecule produced endogenously, is a byproduct derived from HO-1 induction (58, 59). Although toxic at high doses, low-dose CO administered exogenously is cytoprotective in preclinical sepsis and ARDS models (60). CO exerts its protective effect through pleiotropic mechanisms via regulation of autophagy and apoptosis, as well as through regulation of mitochondrial biogenesis and mesenchymal stem/stromal cells (MSCs) (57, 61–63). CO also enhances production and generation of pro-resolving lipid mediators (ref. 64 and see below). A recent multicenter phase I clinical trial examining the role of low-dose inhaled CO in sepsis-induced ARDS showed that low-dose inhaled CO administration in patients with sepsis-induced ARDS was feasible and safe, raising the prospect of a future targeted therapeutic to modify a DAMP-mediated injury cascade (65).

In addition to autophagy, programmed cell death plays an important role in lung inflammation and sepsis-induced lung injury. Previously, apoptosis, which is considered noninflammatory, was believed to be the only form of programmed cell death (66). More recently, other forms of programmed cell death have been reported, including necroptosis (necrotic cell death; ref. 67) and pyroptosis (68), both of which are proinflammatory. Pyroptosis is a caspase-1–dependent form of programmed cell death that occurs in response to infection with intracellular pathogens and is important for pathogen clearance. Immune cells recognize foreign pathogens, release proinflammatory cytokines, then “burst” and die, promoting tissue inflammation. Pyroptosis activated by the inflammasome can promote lung injury (69), including that caused by sepsis (51). Necroptosis is regulated by activation of the RIPK1-RIPK3 kinase complex pathway. Subsequent phosphorylation of the mixed-lineage kinase domain-like protein by receptor-interacting protein kinase 3 (RIPK3) activates necroptosis and leads to the release of cellular contents that propagate the inflammatory response (69). Many of the details of necroptosis signaling have been delineated in cancer cell lines, though necroptosis has recently been shown to contribute to VILI (70). RIPK3-deficient mice were protected from VILI and exhibited lower indices of cell necrosis compared with control animals. Furthermore, RIPK3 was found to be elevated in the plasma of mechanically ventilated patients, many of whom had sepsis-induced ARDS, demonstrating a role for this pathway in critically ill patients (70). In a subsequent study, elevated RIPK3 levels in plasma were associated with in-hospital mortality and organ failure in 5 ICU cohorts in the US and Korea (71). This work makes up part of a rapidly expanding network of nonapoptotic cell death pathways that may also contribute to organ dysfunction in sepsis-induced lung injury (67). Therapies that modulate these cell death pathways may limit inflammation or barrier disruption in the context of sepsis-induced ARDS, and cell death pathways may represent therapeutic targets in lung injury.

Dysregulated inflammation and downstream effectors. Dysregulation of the inflammatory response in sepsis and ARDS has been reviewed previously (30, 40, 41). We will highlight some of the key findings, including recent progress using inflammatory profiles in patient phenotyping. Activation of the innate immune system by PAMPs or DAMPs leads to the release of numerous chemokines, as well as proinflammatory cytokines (e.g., IL-1β, TNF, IL-6, IL-8) and antiinflammatory cytokines (e.g., IL-10) (30). Proinflammatory cytokines are important for pathogen clearance, but high levels can lead to alveolar-capillary barrier breakdown by injury to the endothelial layer (circulating cytokines) or injury to the epithelial layer (alveolar fluid cytokines) (72). After disruption of the alveolar-capillary barrier, proinflammatory effectors in the alveolar fluid are released into the circulation to promote further inflammation and immune responses. Notably, the use of lung-protective ventilation (i.e., low–tidal volume ventilation) has been shown to mitigate the release of various inflammatory cytokines (73).

Recently, it has become clear that there are subphenotypes of ARDS. Calfee et al. identified a hyperinflammatory subphenotype by applying latent class modeling to 2 cohorts from National Heart, Lung, and Blood Institute (NHLBI) ARDS randomized controlled trials (74). Compared with other ARDS patients, the hyperinflammatory subphenotype is characterized by higher plasma levels of proinflammatory biomarkers, including IL-8 and soluble TNF receptor 1; lower serum bicarbonate; and higher prevalence of sepsis. The hyperinflammatory group had worse outcomes, including higher mortality and fewer ventilator- and organ failure–free days. Moreover, patients with the hyperinflammatory subphenotype responded differently to the application of high positive end expiratory pressure on the ventilator, with a reduction in mortality that was not seen across the entire cohort (74). There may be additional subphenotypes of ARDS, such as a profibrotic subphenotype in which exacerbation of underlying interstitial lung abnormalities mimics ARDS (22).

Neutrophils are key effectors of the innate immune response to sepsis. These cells promote pathogen clearance by engulfing and killing bacteria and by releasing NETs, which are networks of extracellular DNA, including nuclear chromatin with associated proteins, that may facilitate pathogen clearance (Figure 1). However, delayed neutrophil apoptosis and prolonged neutrophilic inflammation can mediate further alveolar-capillary barrier disruption (75, 76). Interestingly, degradation of NETs with DNase1 improves lung injury and mortality in mouse models (77). Additionally, intravascular neutrophil and platelet activation can result in formation of neutrophil–platelet aggregates, leading to secondary capture of neutrophils, further promoting endothelial cell activation and barrier disruption. In preclinical ARDS models, blocking neutrophil–platelet aggregates mitigated the development of ARDS (24, 78). The complement system is also activated during sepsis, and PAMP or DAMP exposure leads to the production of peptides such as C3a and C5a. C5a is a potent proinflammatory peptide that functions as a chemoattractant for leukocytes and an amplifier of the inflammatory response. C5a triggers an oxidative burst in neutrophils, resulting in production of ROS and granular enzymes that contribute to tissue damage and barrier dysfunction (79). C5a can also promote NET formation (80).

Inflammatory network activation also affects the vascular and lymphatic endothelium, resulting in increased expression of adhesion molecules that facilitate binding of leukocytes to the endothelium. Severe sepsis can activate the coagulation system, with increased expression of procoagulant proteins (e.g., tissue factor, platelet activating factor) and a reduction in anticoagulant proteins (e.g., tissue factor pathway inhibitor, activated protein C). Fibrin removal is impeded, in part, by increased plasminogen activator inhibitor type 1 that inhibits the endogenous fibrinolytic system. As a result, increased microvascular clots can form that can promote tissue injury and end-organ dysfunction through impaired perfusion. In more severe cases of sepsis, widespread microvascular thrombosis can lead to disseminated intravascular coagulation and, paradoxically, thrombocytopenia and uncontrolled bleeding, likely from consumption of platelets and clotting factors. The coagulopathy of sepsis has been reviewed in detail previously (81).

Microvascular thrombosis and formation of platelet–leukocyte aggregates are likely key processes mediating sepsis-induced lung injury. Abdulnour et al. recently examined peripheral blood leukocyte number and activation in patients at risk for ARDS included in the Lung Injury Prevention Study with Aspirin trial of aspirin versus placebo (82), where aspirin did not reduce the risk of ARDS at 7 days. Many enrolled subjects had sepsis as their risk factor for ARDS development (>76%). Unexpectedly, biomarkers of intravascular monocyte activation (i.e., monocyte–platelet aggregates) in at-risk patients were associated with ARDS development, pointing toward a role for activated monocytes in early ARDS pathogenesis (83). Future studies will need to address the functional implications of monocyte–platelet aggregates and monocyte subsets in lung injury development.

Immune suppression and immunoparalysis. Suppression of the adaptive immune response in sepsis may lead to a persistent compensatory antiinflammatory phase termed “immunoparalysis.” Following initial cytokine-mediated hyperinflammation, immunoparalysis can lead to nosocomial infections and increased mortality in the later phase of sepsis (84). Although resuscitation strategies have improved early mortality in sepsis, patients often die in the later immunoparalysis phase, which is characterized by organ dysfunction, including the development of VILI in patients who require mechanical ventilation. Reactivation of viral infections, such as CMV, during this period (even in previously immunocompetent hosts) may confer additional risk for ARDS development (85). Strategies to prevent reactivation of latent viral infections is a potential approach to combat immunosuppression in sepsis. A recent phase II study examined whether ganciclovir in CMV-seropositive adults with sepsis-associated respiratory failure would prevent reactivation (86). There was no significant difference in the primary outcome (change in plasma IL-6 levels), but secondary analyses showed that ganciclovir decreased CMV reactivation and duration of mechanical ventilation and increased ventilator-free days. Although current data do not support the use of antiviral therapy to prevent reactivation or secondary infection, future studies will be needed to determine if this type of chemoprophylaxis can improve outcomes in the ICU.

Apoptosis-mediated depletion of CD4+ T cells and B cells has been proposed to lead to sepsis-induced immunoparalysis (87). Another potential mechanism is sepsis-related T cell exhaustion, during which dysfunctional T cells express programmed cell death protein 1 (PD-1) and cytotoxic T lymphocyte antigen-4 (CTLA-4). Stromal and professional antigen-presenting cells increase expression of T cell programmed death ligand 1 (PD-L1) that binds PD-1 on T cells and suppresses T cell function (88). Furthermore, there is a shift toward a T helper type 2 immune-suppressive cytokine profile and increased activity of Tregs, which are also immune suppressive (88). Interestingly, T helper type 17 cells link the adaptive and innate immune systems, in part through IL-17 production and augmentation of neutrophil responses (89). Similar to cancer, reversal of T cell exhaustion has been proposed as a potential therapeutic approach in sepsis (90), and antibodies to CTLA-4, PD-1, and PD-L1 have shown benefit in murine sepsis models (91, 92). Other effectors of the immune system, including NKT cells, mucosal-associated invariant T cells, γδ T cells, CD8+ T cells, and B cells, have previously been implicated in sepsis (see ref. 30 for detailed review).

Another potential strategy to prevent secondary infection and organ dysfunction during sepsis is to target immunodysfunction, with the goal of determining an individual’s state of inflammation and/or immunoparalysis. This approach requires the ability to define the immunologic state of a patient with sepsis in real time and then administer treatment to alter that state. Such phenotyping would determine whether targeted therapy to suppress or boost the immune system could mitigate lung injury, and novel tools for rapid bedside immunophenotyping are being developed (93). Treatment with an antiinflammatory therapy may be appropriate for a patient in the hyperinflammatory phase of sepsis, while immunostimulatory therapies have a theoretical benefit during immunoparalysis. Examples of potential immunostimulatory therapies include leukocyte growth factors (GM-CSFs) that can increase neutrophil phagocytosis and killing (94); immunostimulatory cytokines (IFN-γ and IL-7) that stimulate T cell survival and proliferation (95); and inhibitors of negative costimulatory pathways (PD-1 and PD-L1) (88). A recent study by Morrell et al. suggested that PD-L1 activation might hold promise for patients with sepsis-induced ARDS. The authors used cytometry by TOF (CyTOF) (96), combining flow cytometry with elemental mass spectrometry, to characterize immunophenotypic profiles of human alveolar leukocytes in ARDS (97). The majority of subjects with ARDS in this study had sepsis-induced ARDS, and expression of PD-1– and PD-L1–associated genes was significantly decreased in alveolar macrophages from ARDS subjects who required prolonged mechanical ventilation or died. The findings in this small cohort raise the interesting question as to whether PD-L1 expression on alveolar macrophages plays an important role in controlling the inflammatory response and whether activating PD-L1 may represent a novel therapeutic strategy in ARDS. Interestingly, PD-L1 checkpoint blockade in cancer patients has been associated with development of an ARDS-like pneumonitis (98), further supporting the relevance of this pathway in ARDS.