The acute respiratory distress syndrome

Introduction

Since the original description of the acute respiratory distress syndrome (ARDS) in 1967, considerable progress has been made in understanding the pathogenesis and pathophysiology of acute lung injury (ALI) (1–4). The likelihood of survival is determined by the severity of lung injury, the extent of nonpulmonary organ dysfunction, preexisting medical conditions, and the quality of supportive care. Because ARDS is a complex syndrome with a broad clinical phenotype, it has been challenging to translate the results of cell and animal studies to pharmacologic therapies that reduce mortality in humans. Nevertheless, laboratory-based investigations have produced valuable insights into the mechanisms responsible for the pathogenesis and resolution of lung injury, and preclinical studies paved the way for important improvements in supportive care. Two of these therapies, lung-protective ventilation and fluid-conservative management, have reduced mortality and morbidity, respectively. This review of ARDS will focus on some of these issues, including new insights into the molecular mechanisms of lung injury and repair.

Definitions, epidemiology, incidence, and mortality

Criteria for the diagnosis of ARDS have evolved. The original description emphasized rapidly progressive respiratory failure from noncardiogenic pulmonary edema, requiring mechanical ventilation because of severe arterial hypoxemia and difficulty breathing (5). In 1988, a 4-point scoring system provided a quantitative assessment of lung injury severity based on the degree of hypoxemia, the level of positive end-expiratory pressure (PEEP), static respiratory compliance, and the extent of radiographic infiltrates (6), and this scoring system has been useful for research and clinical trials. In 1994, a consensus conference recommended simplified criteria: arterial hypoxemia with PaO₂/FiO₂ ratio less than 300 mmHg and less than 200 mmHg to define ALI and ARDS, respectively, and bilateral radiographic opacities without evidence of left atrial hypertension (7). These criteria have been widely utilized, although some investigators believe that the definitions should specify the level of PEEP and/or the fraction of inspired oxygen. A recent report — what is now called the Berlin definition — recommends use of three categories of ARDS, based on the degree of hypoxemia: mild (200 mmHg < PaO₂/FiO₂ ≤ 300 mmHg), moderate (100 mmHg < PaO₂/FiO₂ ≤ 200 mmHg), and severe (PaO₂/FiO₂ ≤ 100 mmHg) (8). Whether stratification of patients based on these descriptions will advance the efficacy of clinical detection and of charting the natural history of ARDS remains to be determined.

Most investigations have focused on ALI and/or ARDS patients who are already mechanically ventilated. Recently, progress has been made in diagnosing ALI in spontaneously breathing patients who have bilateral pulmonary infiltrates and arterial hypoxemia and in whom intravascular volume overload and congestive heart failure are excluded (9, 10). This approach facilitates patient identification and testing of new therapies prior to the need for mechanical ventilation. Figure 1 provides a clinical vignette describing early recognition of ALI.

Bacterial or viral pneumonia is the most common cause of ALI and ARDS (1). Sepsis due to nonpulmonary infections, aspiration of gastric contents, and major trauma with shock also commonly precipitate the injury. Less commonly, acute pancrea­titis, transfusions, drug reactions, and fungal and parasitic lung infections are linked to ALI and ARDS. The coexistence of two or more of these risk factors can enhance the likelihood of developing ALI or ARDS (1).

A prospective epidemiologic study in 1999–2000 estimated an annual incidence of ALI and ARDS of 190,000 adult patients in the United States (11). There is a substantial incidence of ALI and ARDS in children as well (12, 13). Data from 2001–2008 indicate that the incidence of ALI and/or ARDS in hospitalized adults has declined, perhaps secondary to more widespread use of lung-protective ventilation, reductions in nosocomial infections, and more conservative use of blood products (14, 15).

Severe arterial hypoxemia (PaO₂/FiO₂ < 100) and an increase in the pulmonary dead space fraction (>0.60) are associated with higher mortality (16), as are shock, liver dysfunction, acute kidney injury, age over 60 years, and higher severity of illness scores (17–19). Based on the NIH Heart, Lung and Blood Institute (NHLBI) ARDS Network trials, 60-day mortality has declined from 36% in 1996–1997 to 26% in 2004–2005 (20). The most recent ARDS Network clinical trials reported a 60-day mortality of 22% in adult patients despite higher APACHE III scores and a higher incidence of shock at enrollment compared with a prior trial in 2006 (Figure 2 and ref. 4).

Environmental and genetic influences

Environmental and genetic factors that contribute to susceptibility and severity of ALI and ARDS have emerged as a major research focus. Chronic alcohol abuse increases the risk of ALI and ARDS (21) and multiple organ failure in septic shock (22). Both active and passive cigarette smoke exposure, as quantified by plasma levels of cotinine, have been independently associated with the development of ALI after severe blunt trauma (23). The mechanisms may include priming effects of cigarette smoke exposure on the lung endothelium, alveolar epithelium, or inflammatory cells.

Variants in more than 25 genes have been associated with developing ALI and/or ARDS and with clinical outcomes (24) including common variants of genes that regulate inflammation, coagulation, endothelial cell function, reactive oxygen radical generation, and apoptosis (25–29) — all processes that are important in lung injury and repair (2, 30, 31). For example, the Fas pathway modulates apoptosis, inflammation, and epithelial cell injury; in a candidate gene study, common genetic variants in Fas were associated with susceptibility to developing clinical lung injury (27). African Americans with ALI have a higher risk of death compared with white patients. A candidate gene study identified a functional T-46C polymorphism (rs2814778) in the promoter region of the Duffy antigen/receptor for chemokines (DARC) gene that was associated with a 17% increase in 60-day mortality in African-American patients enrolled in ARDS Network clinical trials. Plasma interleukin-8 levels were increased in those individuals with the DARC polymorphism, supporting one mechanism contributing to a worse clinical outcome (29).

Genetic polymorphisms that predispose individuals to the injurious effects of specific bacteria or viruses may influence the development of ALI and ARDS. Indeed, several polymorphisms are associated with more severe pneumococcal, Legionella, and viral lung infections (32). The genetic factors that regulate the virulence of infecting pathogens also require more research (33) to relate the severity of clinical lung injury to specific microbiologic variables that contribute to severe pneumonia and ALI. Genetic features of both the host and the microbe likely are important in determining the severity of lung injury.

In spite of the intriguing associations, putative causal genes require more validation in independent study populations, along with studies of gene and environmental interactions that may alter these associations. Genome-wide association studies for susceptibility and outcomes in multiple populations are in progress. One recent study of this nature identified PPFIA1 (which encodes liprin α, a protein involved in cell adhesion, integrin expression, and cell-matrix interactions) as a predictor for developing acute lung injury after major trauma (25).

Pathogenesis: dysregulated inflammation and alveolar barrier disruption

Dysregulated inflammation, inappropriate accumulation and activity of leukocytes and platelets, uncontrolled activation of coagulation pathways, and altered permeability of alveolar endothelial and epithelial barriers remain central pathophysiologic concepts in ALI and ARDS (1–3). Activation of the innate immune response by binding of microbial products or cell injury–associated endogenous molecules (danger-associated molecular patterns [DAMPs]) to pattern recognition receptors such as the Toll-like receptors on the lung epithelium and alveolar macrophages is now recognized as a potent driving force for acute lung inflammation (34). Newly reported innate immune effector mechanisms, such as formation of neutrophil extracellular traps — lattices of chromatin and antimicrobial factors that capture pathogens but also can cause endothelial injury — and histone release by neutrophils (35) may contribute to alveolar injury. Signaling between inflammatory and hemostatic effector cells, such as platelet-neutrophil interaction, is important in some models, including acid-induced ALI, sepsis, and transfusion injury (36, 37). The delicate balance between protective and injurious innate and adaptive immune responses and hemostatic pathways may determine whether alveolar injury continues or is repaired and resolved. For example, in lung infection, acute inflammatory responses to pathogens and their toxins (38–40) cause ALI through leukocyte protease release, generation of reactive oxygen species, rampant synthesis of chemokines and cytokines, Toll-like receptor engagement, and actions of lipid mediators (33, 41, 42). Nevertheless, these same inflammatory mechanisms, when controlled rather than excessive, are requisite in pathogen containment and clearance. Recent research suggests that other pathways, such as the molecular events that govern the balance between angiotensin converting enzymes 1 and 2, may influence the degree of inflammatory lung injury consequent to viral infection and sepsis (43, 44). Similarly, newly recognized lipid modifications may contribute to resolution of lung inflammation (45).

Increased permeability of microvascular barriers, resulting in extravascular accumulation of protein-rich edema fluid, is a cardinal feature of acute inflammation and a central pathophysiologic mechanism in ALI and ARDS (Figure 3A and refs. 1, 3, 46). Increased permeability is also linked to transfer of leukocytes and erythrocytes into the alveolar space in ARDS (46), as well as to inflammasome-regulated cytokines (47). A variety of mediators, pathways, and molecular systems contribute to altered alveolar endothelial and epithelial permeability (48–53). Vascular endothelial cadherin (VE-cadherin), an adherens junction protein, is critical for maintenance of endothelial barrier integrity in lung microvessels (54). Disruption of VE-cadherin homophilic bonds destabilizes lung microvascular barrier function (Figure 3B). Antibodies against VE-cadherin, destabilizing agonists such as TNF, thrombin, and VEGF, and leukocyte signals all interrupt VE-cadherin bonds and induce lung edema formation (54, 55). In contrast, stabilization of VE-cadherin bonds (Figure 3C) through genetic manipulation of VE-cadherin–catenin interactions (56) or prevention of dissociation of a phosphatase from VE-cadherin (57) reduces BAL protein and leukocytes in LPS-challenged mice. Thus, experimental manipulation of VE-cadherin alters alveolar and systemic endothelial barrier function and leukocyte transmigration, with pathogenic implications for clinical ALI and ARDS.

Molecular approaches to specifically reverse increased-permeability pulmonary edema have been long sought in ALI and ARDS research (1, 3). Candidate pathways for stabilization of lung and systemic endothelial barriers (reviewed in ref. 58) have recently been described. Systemic endothelium may be a critical therapeutic target in septic ALI (58) and in multiple organ failure associated with ALI and ARDS (1). A central therapeutic paradigm involves administration of stabilizing ligands that bind to receptors on endothelial cells and activate intracellular pathways, mediating cytoskeletal reorganization and catenin–VE-cadherin interactions that tighten VE-cadherin bonds (Figure 3C). Sphingosine-1-phosphate (S1P) is a lipid recognized by G protein–coupled receptors on endothelial cells (S1Pr1, S1Pr2, S1Pr3). S1P binding to S1Pr1 induces actin cytoskeletal reorganization, RAC activation, localization of α-, β-, and γ-catenin and VE-cadherin to regions of intercellular contact, and assembly of adherens junctions in cultured human endothelial cells (59, 60). S1P enhances pulmonary and systemic endothelial barrier integrity in vivo and in vitro (reviewed in ref. 61), and small-molecule agonists of endothelial S1Pr₁ suppress cytokine storm and lung leukocyte recruitment in experimental influenza (62). S1P is present in high concentrations in plasma and in that compartment regulates basal and inflammation-triggered vascular leak in the lungs and systemic vessels of mice (63). Platelets may locally contribute S1P at sites of vascular injury (60) and may reduce alveolar hemorrhage — another complication of endothelial barrier disruptions (46) — under some conditions; this effect may in part be related to delivery of S1P. However, effects of S1P or synthetic S1P receptor agonists may depend on biologic context (60) and time/duration of administration (64), since S1Pr₂ and S1Pr₃ are barrier destabilizing and S1Pr₁ undergoes time-dependent desensitization (60, 63). Similarly, other receptor-mediated signaling systems, such as those recognized by thrombin and other G protein–coupled protease-activated receptor agonists, may differentially trigger lung endothelial barrier disruption or stabilization, depending on time and context (65).

The Robo4/Slit signaling system also stabilizes the endothelial barrier (66, 67). In contrast to S1P receptors (60), Robo4 expression is restricted to endothelial cells (68). An active fragment of the Robo4 ligand Slit (Slit2N) inhibits tyrosine phosphorylation of VE-cadherin and preserves its association with P120 catenin, preventing VE-cadherin internalization and abnormal permeability of human microvascular endothelial cells induced by TNF-α, IL-I, or LPS (67). In mice, Slit2N reduces pulmonary and systemic vascular permeability in LPS lung injury, cecal ligation and puncture, and influenza infection, increasing survival (67). Cytokine levels are not decreased, indicating that Robo4 signaling does not inhibit this component of inflammation and that endothelial barrier stabilization may be sufficient to improve outcomes in lethal infectious challenges (58, 67).

Although numerous endothelial-stabilizing agonists and intracellular pathways have been identified (refs. 58, 61, and Figure 3C), use of these agents may have unintended consequences. Molecular mechanisms by which plasminogen activator inhibitor-1 mediates Pseudomonas-associated alveolar endothelial barrier disruption have been identified, but Pai1^–/– mice have a defect in alveolar neutrophil recruitment and increased mortality compared with wild-type animals (69). This emphasizes the daedal relationships between barrier integrity and leukocyte transmigration and the precarious tension between injurious and protective inflammatory mechanisms that may operate in ALI and ARDS.

In contrast to endothelium, less is known about the potential mechanisms of alveolar epithelial stabilization, although epithelial permeability is critical in alveolar flooding (70, 71) and leukocyte accumulation (52) and potentially critical for intra-alveolar fibrin deposition and hyaline membrane formation (72). Epithelial barriers involve cadherin-mediated adherens junction bonds and tight junctions, although the topography differs from endothelium, and E-cadherin substitutes for VE-cadherin (54). Alveolar epithelial barriers are tighter than alveolar endothelial barriers (70), but the two have functional interactions (2). For example, under some (55), but not all (57), conditions, disruption of endothelial VE-cadherin bonds causes alveolar epithelial leak and epithelial injury. Mesenchymal stem cells (MSCs) (see Future directions) restore barrier integrity in cytokine-treated cultured human alveolar epithelial cells (73). The mechanism involves release of angiopoietin-1, which inhibits actin stress fiber formation and redistribution of the tight junction protein claudin 18 in epithelial cells (73) and also induces S1P production and inhibits endothelial VE-cadherin internalization (58).

Resolution

Resolution of ALI requires effective and synchronous (a) reabsorption of alveolar edema, (b) repair of the epithelial and endothelial barriers, and (c) removal of inflammatory cells and exudate from the distal airspaces (Figure 4). Resolution of ALI and repair of alveolar structures may — like the initiation of alveolar damage — depend on a precise balance of inflammatory interactions and molecular signaling. For example, hyaluronan fragments found in the serum of ALI patients trigger release of chemokines by macrophages but also interact with Toll-like receptors to deliver signals that limit epithelial apoptosis and promote reestablishment of epithelial integrity in experimental lung injury (74).

Reabsorption of alveolar edema (Figure 4A) occurs through vectorial transport of sodium and chloride across alveolar epithelial type I and II cells to create a mini–osmotic gradient to reabsorb water (75–77). This process is impaired in ALI and ARDS because of apoptosis and necrosis of alveolar epithelium (31, 46) and defects in transcellular ion transport induced by proinflammatory cytokines, oxidants, and hypoxia (75, 78–81). Effective reabsorption of edema fluid from the air spaces requires reestablishment of an effective epithelial barrier (70, 82). Reepithelialization occurs initially by proliferation of type II cells (46), and the traditional view is that type II cells are the main source of new alveolar epithelial cells (83). Nevertheless, there is new evidence that progenitor cells may exist in strategic niches in bronchoalveolar junctions and that the alveolar epithelium can be activated to regenerate the epithelial and endothelial barriers (84–87). For example, α6β4-expressing progenitor cells have been identified that account for a substantial fraction of the type II cells that reepithelialize the injured alveolar barrier (Figure 4C and ref. 88). There may also be a human c-kit⁺ lung stem cell capable of renewing all lung cells (89), although this hypothesis is controversial and requires validation. There has been progress in understanding how inflammation is resolved through clearance of alveolar neutrophils, monocytes, and necrotic debris including the contributions of lipid mediators such as lipoxin A4, resolvin E1, and other antiinflammatory pathways (Figure 4B and refs. 45, 90). Macrophages remove apoptotic neutrophils and monocytes via molecular mechanisms that have recently been more clearly identified (91–93). A deficiency of alveolar macrophages worsens influenza-related pneumonia and lung injury in mice, leading to an increase in neutrophils and neutrophil extracellular traps (94). Lymphocytes also contribute to resolution. In a mouse model of endotoxin-induced lung injury, CD4⁺CD25⁺ regulatory T cells that were recruited to the air spaces played an essential role in resolving inflammation by enhancing neutrophil apoptosis and suppressing cytokine secretion, in part by release of TGF-β (95).

New insights based on clinical trials

Many clinical trials have assessed pharmacologic interventions, innovative strategies for positive-pressure ventilation, and other supportive approaches to ALI and ARDS treatment, advancing our understanding of the mechanism of the disease (Table 1).

Future directions

Cell-based therapy with allogeneic human MSCs has emerged as a promising approach to therapy for ALI and ARDS (139–144). MSCs secrete multiple effector molecules, including antiinflammatory cytokines, growth factors, and antimicrobial peptides. These can reverse the major abnormalities of lung injury, including altered lung endothelial and epithelial permeability, impaired alveolar edema fluid clearance, dysregulated inflammation, and infection. In addition, in an experimental model, MSCs attached to the alveolar wall by connexin-43–based gap junctional channels and transferred mitochondria to endotoxin-injured alveolar epithelium, restoring alveolar ATP production and normalizing surfactant production and epithelial barrier properties (145). Also, the discovery that short-term function can be achieved in repopulated decellularized lungs raises the prospect that basic understanding of lung regeneration (146), including signals from the lung matrix, may lead to novel strategies to enhance repair in the severely injured lung.

With declining mortality, clinical trials of new therapies will need to enroll patients with the highest predicted mortality (4) and include approaches that combine clinical and biological predictors (147, 148) to select the patients at highest risk. Clinical trials are in progress to test whether statin therapy can favorably modulate inflammation and restore barrier integrity in ALI and ARDS (Figure 3C). In animals, aspirin was effective in preventing acid-induced and transfusion-associated lung injury (37). Consequently, a clinical trial is planned to test the efficacy of aspirin in patients at high risk for developing ALI. There is also progress on developing more effective extracorporeal systems to support gas exchange for patients with very severe lung injury (149). Further research into the complex factors that influence the functional status of patients after recovery from ALI or ARDS (150) may also stimulate new approaches to attenuate long-term muscle and neurologic impairments that cause disabilities in ALI and ARDS survivors.