Chapter 37. Ventilator-Induced Lung Injury

• Ventilator-induced lung injury may occur with both lung volumes that lead to overdistention of lung units (volutrauma) or with low distending pressures that lead to the development of atelectasis (atelectrauma).
• Ventilator-induced lung injury may cause injury in previously healthy regions of lung, and may also lead to multiorgan dysfunction.
• To reduce the risk of ventilator-induced lung injury, low tidal volumes of 6 mL/kg should be used in treating most patients with acute respiratory distress syndrome.
• The appropriate level of positive end-expiratory pressure remains to be determined, but levels of PEEP that minimize atelectrauma may be beneficial.
• Permissive hypoventilation (hypercapnia) may be a necessary component of a lung-protective ventilatory strategy.

There is consistent and convincing experimental evidence that mechanical ventilation, particularly in the setting of lung injury, can contribute to functional and structural alterations in the lung. The experimental evidence has led to the notion that mechanical ventilation not only perpetuates the lung injury, but also contributes to both the morbidity and mortality of the acute respiratory distress syndrome (ARDS). Concern surrounding ventilator-induced lung injury (VILI) culminated in a consensus conference in 1993 that made (based solely on studies in animal models of ARDS) the empirical recommendation to limit tidal volumes to the range of 5 to 7 mL/kg and plateau pressures less than 35 cm H2O.1 It would be 8 years until the recommendations of the consensus group were affirmed by a randomized controlled trial demonstrating that a lung-protective strategy designed to limit VILI would lead to an improvement in patient outcome.2 Unfortunately, it seems that it may take even longer until there is incorporation of these concepts into widespread clinical practice.3

The objectives of this chapter are to develop the concept of VILI and provide the rationale for the shift in ventilation philosophy for patients with ARDS. First, the relevant features of ARDS as it pertains to VILI will be reviewed, since most of the studies evaluating VILI have focused on ARDS. Then, the concept of lung-protective ventilation strategies will be discussed, and pertinent studies evaluating these newer strategies in patients with ARDS will be presented. Recommendations based on current clinical evidence, and when this is lacking best experimental evidence, will also be presented (Table 37-1).

ARDS is characterized by endothelial and epithelial cellular injury. This loss of integrity of the alveolar-capillary membrane results in a high-permeability pulmonary edema and formation of hyaline membranes. Injury to type II pneumocytes also occurs, and there are alterations in surfactant function. Plain chest radiographs are frequently misleading, in that they suggest that the lung damage is uniformly distributed. However, evaluation of patients with ARDS using computed tomography (CT) of the thorax has demonstrated that the airspace disease is patchy.4,5 Marked heterogeneity and regional differences in lung injury exist in the lungs of these patients. Regions of lung with airspace disease are contrasted to adjacent areas with normal-appearing alveoli. In addition, an exaggerated vertical gradient of lung inflation has been demonstrated in ARDS, with compression of alveoli and a decrease in aerated lung as one progresses from nondependent to more dependent lung regions.6 Indeed as emphasized by Mead and coworkers in the early 1970s it is likely this degree of heterogeneity of the lung injury that makes the lung uniquely susceptible to the effects of ventilator-induced injury.7 The heterogeneity of the lung injury is also responsible for the decrease in lung compliance that characterizes ARDS. It is worth emphasizing that this loss of compliance is due to a functional reduction in alveolar units and not due to the development of “stiff” lungs. Indeed the recognition that ARDS is characterized by a loss of functional lung units with preservation of other alveoli with normal lung compliance (specific lung compliance) is central to the current notion of lung protection strategies.8 In many respects owing to the reduction in effective lung volume, the 70-kg adult patient with ARDS must be treated from the pulmonary point of view as a 30-kg pediatric patient. Consequently the use of traditional tidal volumes of 10 to 15 mL/kg (700 to 900 mL in our 70-kg patient) would be inappropriate and will simply result in overdistention of lung units with relatively normal compliance. The need to modify the approach to mechanical ventilation in ARDS is further emphasized by three decades of investigations that demonstrate that the overdistention of lung units may in itself lead to lung injury identical to that seen in ARDS. ARDS is also a syndrome characterized by inflammation of the lung with various cytokines and other mediators thought to play a major role. In recent years, there has been a large body of evidence indicating that mechanical ventilation may have an impact on this aspect of the pathophysiology of ARDS, and indeed there is the suggestion that the improvement in mortality with lung-protective strategies may be partly due to a reduction in release of various mediators by these strategies.

Macroscopic Injury

Recent evidence suggests that mechanical ventilation may have both regional and systemic effects. VILI may be broadly classified into macroscopic and microscopic injury (Table 37-2). Macroscopic injury consists of what has been classically described as barotrauma. Pneumothorax, pneumomediastinum, pneumoperitoneum, and subcutaneous emphysema are recognized complications of mechanical ventilation, and are characterized by the presence of extra-alveolar air.9 The mechanism of the development of extra-alveolar air was elegantly demonstrated by Macklin and Macklin in 1944.10 During positive-pressure ventilation (PPV), alveoli rupture at the point where the alveolar base meets the bronchovascular sheath. Air then dissects along the vascular sheaths towards the mediastinum and into the hilum and mediastinal soft tissues. If sufficient gas accumulates, the mediastinal parietal pleura ruptures, and a pneumothorax develops.

Macroscopic barotrauma correlates with a variety of factors. In a retrospective study in 139 intubated patients, barotrauma occurred in 34 patients.11 Peak airway pressure, level of positive end-expiratory pressure (PEEP), tidal volume, and minute ventilation correlated with the development of barotrauma. However, in a subsequent prospective study of 168 patients over a 1-year period, only the presence of ARDS was associated with the development of barotrauma.12 Gas trapping has also been associated with the development of extra-alveolar air leakage. In a study in mechanically ventilated asthmatics, an elevation in end-inspiratory lung volume above 1.4 L was associated with macroscopic barotrauma.13 The relationship of PEEP to the development of extra-alveolar air is inconsistent.12,14 Patients with severe underlying lung disease often require higher levels of PEEP to maintain oxygenation, and it is possible that it is the underlying lung disease in such patients that explains the correlation between air leaks and PEEP levels. More recently Eisner and colleagues, using data from the ARDS Network trial, reported that higher PEEP was associated with an increased risk of barotrauma (relative risk = 1.5; 95% confidence interval [CI] 0.98 to 2.3).15 The strength of the association in this trial was the ability to control for other potentially relevant risk factors and uniformity of the ventilation methodology. One will have to see how this observation is born out by clinical trials that are currently underway to evaluate higher PEEP strategies in patients with ARDS. Patients with underlying lung damage are particularly prone to the development of macroscopic barotrauma. Asthma, chronic obstructive pulmonary disease (COPD), and pneumonia have all been identified as risk factors.11 ARDS has consistently been associated with a very high risk for the development of extra-alveolar air, with an incidence in some series of 40%.11,12,16 In addition to the immediate hemodynamic consequences, the development of macroscopic barotrauma in patients receiving mechanical ventilation portends a worse prognosis. CT studies of patients with ARDS have expanded the scope of macroscopic injury. Gattinoni and coworkers have described the appearance of bullae and cystic parenchymal lesions located predominantly in the dependent (dorsal) lung regions.4 These lesions are often occult and are not readily detected on plain chest radiographs. In a histologic study in patients with severe respiratory failure, airspace enlargement with alveolar distention in aerated lungs or intraparenchymal pseudocysts in nonaerated regions were identified in 87% of the lungs examined. Importantly, patients with severe airspace enlargement were ventilated with higher airway pressures and higher levels of inspired oxygen and had a higher rate of pneumothoraces. One reasonable hypothesis is that the cystic changes observed in aerated lungs occur because of the high airway pressures causing overdistention of alveoli. The etiology of the cystic lesions observed in nonaerated, more dependent lung regions remains speculative. Since they occur in dependent regions of the lung, their appearance is not explained by lung overdistention. Possibly they are caused by repetitive opening and closing of nonrecruited alveoli (see below). Alternatively, because perfusion is greater in dependent regions, the cystic lesions may be caused by blood-borne factors such as cytokines and inflammatory cells.

The importance of lung perfusion was emphasized by Broccard and coworkers, who used an isolated rabbit lung preparation and demonstrated that an increase in lung perfusion lead to more severe hemorrhage, an increase in filtration coefficient, and heavier lungs.17 Injury to conducting airways could also potentially lead to an increase in regional airways resistance, with resultant gas trapping and progressive downstream regional lung distention. Regions of local superinfection and resultant inflammation may intensify the bronchiolar injury. Goldstein and associates used a piglet model and found cystic lung changes and areas of bronchiolectasis in animals that received intrabronchial inoculation with Escherichia coli.18 The importance of bronchiolectasis in the pathogenesis of VILI is further highlighted by observations that the dead space (a potential prognostic marker in ARDS19) correlated with the presence and severity of bronchiolar injury and dilation.20 In summary, macroscopic lung injury represents a continuum from airspace enlargement through interstitial emphysema and eventually to radiographically apparent extra-alveolar air.

Microscopic Injury

More recent investigations have centered on the development of microscopic lung injury related to mechanical ventilation. Shortly after the institution of invasive positive pressure ventilation as a therapeutic modality, the development of lung damage was observed in animals ventilated for prolonged periods. The term “respirator lung” was coined to describe the functional and histologic features.21 Subsequent investigations have demonstrated that mechanical ventilation, even at modest airway pressures, is capable of producing functional impairment of the lung with loss of integrity of the alveolar-capillary barrier, surfactant dysfunction, and parenchymal damage that mimics the histologic appearance of ARDS. These observations have led some investigators to speculate that mechanical ventilation itself could be contributing to some of the lung injury, morbidity, and mortality in patients with acute respiratory failure.

In 1974, Webb and Tierney graphically illustrated the deleterious effects of mechanical ventilation in rats using varying levels of peak airway pressure and PEEP.22 Animals ventilated using low peak airway pressures (14 cm H2O and no PEEP) had no pathologic or physiologic changes. In contrast, rats ventilated with peak pressures of 30 cm H2O and no PEEP had perivascular edema and no alveolar edema. These findings were magnified in rats ventilated with peak pressures of 45 cm H2O. Alveolar and perivascular edema developed, along with severe hypoxemia, decreased dynamic compliance, and obvious gross anatomic changes (Fig. 37-1). Interestingly, rats ventilated using 10 cm H2O of PEEP and peak pressures of 45 cm H2O had no alveolar edema (see Fig. 37-1, center). This latter finding led to the concept of a protective effect of PEEP, which will be discussed later.

High Airway Pressures and Lung Injury

High airway pressures themselves are not responsible for lung injury. As anyone who has seen Dizzy Gillespie play the horn knows, high airway pressures do not produce alveolar disruption. Trumpet players are capable of generating large increases in airway pressure, up to 150 cm H2O, without pulmonary sequelae.23 The key factor causing lung injury is an increase in lung stretch that is best assessed in the pressure domain by the transpulmonary pressure (alveolar minus pleural pressure). For Dizzy Gillespie to generate such high airway pressures, he had to generate a high pleural pressure so the transpulmonary pressure was not elevated. Recently the emphasis has shifted away from the concept of pressure-induced injury (barotrauma) to lung overdistention (volutrauma). In this regard, peak inspiratory pressure (PIP) is often used at the bedside as a surrogate marker for the degree of lung inflation. However, PIP is dependent on the resistive pressure drop arising from flow across the endotracheal tube and conducting airways, and is also dependent on the compliance of the respiratory system. Consequently, PIP may not be indicative of the degree of lung inflation; plateau pressure obtained after occlusion of the airway following inspiration more accurately reflects alveolar pressure, but is also hampered by the compliance of the respiratory system, and thus is also an inaccurate surrogate of lung distension. In normal individuals, full inspiration to total lung capacity (TLC) occurs at a transpulmonary pressure of 20 to 25 cm H2O. Patients with hypoxic respiratory failure, however, are commonly ventilated with very high inspiratory pressures, producing transpulmonary pressures greater than 20 to 25 cm H2O.11 Plateau pressures in excess of 30 cm H2O occur frequently and are often beyond the level expected to produce full inspiration.24 The development of such high airway pressures in patients with ARDS could in turn lead to high transpulmonary pressures, and consequently to overdistention of lung units and lung injury.

Direct evidence for volutrauma stems from several observations in animals. To separate the effects of pressure and changes in lung volume, excursion of the chest (and hence tidal volume) was limited by binding the thorax during PPV.25 Open-chest animals whose lungs were ventilated with peak pressures of 15 cm H2O for 15 minutes had an 850% increase in their filtration coefficient. In contrast, the filtration coefficients of closed-chest animals increased by 31% with a PIP of 30 cm H2O, and by 450% with a PIP of 45 cm H2O. Animals in which lung expansion was limited did not develop lung edema during ventilation with peak airway pressures up to 45 cm H2O. The concept of volutrauma as opposed to a pure pressure-related effect was further supported in a study that compared the development of edema with the use of negative-pressure ventilation and high tidal volumes, to that with the use of PPV to achieve similar tidal volumes.26 The amount of edema obtained using negative-pressure ventilation was comparable to that in the animals ventilated with PPV at the same tidal volumes. These studies are important for two reasons. First, they established the importance of lung inflation, and second they demonstrated a dose-response relationship, with larger transpulmonary pressures and tidal volumes producing an incremental increase in lung injury and edema.27 Theoretically, overdistention of lung units may lead to mechanical disruption of the alveolar-capillary barrier and the development of pulmonary edema.28 In a very illustrative histologic study using electron microscopy, overdistention of the lung increased the number of endothelial and epithelial breaks and contributed to rupture of the blood-gas barrier and separation of the epithelium from its underlying basement membrane (Fig. 37-2).

Patients with patchy regions of lung injury may be particularly susceptible to the development of VILI. A nonuniform distribution of lung injury with regional differences in lung compliance may lead to regional differences in lung inflation. The importance of regional overdistention of lung units was emphasized in a study in dogs that compared the effect of delivering a PIP of 40 cm H2O to the entire lung and delivering the same PIP to an isolated lung segment. Inflation of the entire lung did not result in edema.29 However, distention of a small area for 20 minutes produced a protein-permeable epithelium. Thus inflation limited to a subsection of lung resulted in overdistention as the surrounding lung was compressed and allowed for greater expansion for a given PIP. This observation is particularly germane to the discussion of potential VILI in patients with inhomogeneous lung compliance, such as that described in ARDS. Consequently, the notion of a “safe” level of peak airway pressure in these patients must take into account the heterogeneous nature of the disease.

In addition to a time effect, there appear to be species- and age-specific differences in susceptibility to VILI. Younger and smaller animals seem more susceptible to the injurious effects of high airway pressures.22,29–31 It has been postulated that this finding is due to immaturity of the alveolar-capillary barrier. Importantly, injured lungs also appear to be especially susceptible to the development of VILI. In rabbits with mild oleic acid–induced lung injury, the effects on lung edema of mechanical ventilation using high airway pressures were greater than those of either lung injury or mechanical ventilation alone.32 Consequently, patients with underlying acute lung injury secondary to ARDS or pneumonia may be exquisitely sensitive to the adverse effects of mechanical ventilation, and predisposed to the development of VILI.

Gas flow rate may also be important in the development of lung injury. In one study, high flow rates were associated with an increase in filtration coefficient despite similar peak inspiratory pressures, although end-inspiratory lung volume was not accounted for.33

In summary, there is persuasive experimental evidence to support the concept of lung overdistention as a major component of VILI. Studies have demonstrated a dose-response relationship, in which both higher inflation pressures and tidal volumes administered over progressively longer periods of time produce a graded severity of lung injury. Second, VILI has been demonstrated in many animal models, negating the possibility that these observations were due to a species-specific effect.

The Role of End-Expiratory Lung Volume

In addition to lung overdistention, it is becoming increasingly apparent that ventilation at low lung volumes may also lead to lung injury. Webb and Tierney22 demonstrated that when animals were ventilated using high airway pressures without PEEP, significant edema and lung injury resulted. These effects were not seen when PEEP of 10 cm H2O was used, even though the peak airway pressures were identical (see Fig. 37-1). Subsequent studies have confirmed these findings. For example in a canine aspiration model, low levels of PEEP during mechanical ventilation were associated with an increase in edema compared to ventilation with higher PEEP levels.34 The precise mechanism of the effect of PEEP on VILI is controversial and potentially multifactorial. The application of PEEP is known to have effects on distribution of lung water, pulmonary hemodynamics, and pulmonary compliance. In patients with ARDS, providing PEEP may have one of three effects on the state of lung inflation. First, lung units that are already aerated may show increased distention, which could contribute to VILI. Second, partially aerated lung units or those that collapse at the end of a tidal breath may be kept patent during the entire respiratory cycle. Third, previously closed alveoli may be recruited, leading to an increase in functional lung and resulting in an increase in total lung compliance. The latter two points deserve further emphasis. In addition to an upper inflection point, a lower inflection point has been described in patients with ARDS. With a progressive increase in end-expiratory pressure, there is a gradual increase in end-expiratory lung volume or functional residual capacity (FRC), until a point is reached at which compliance improves.35,36 This finding was supported in studies using CT, in which PEEP was found to reduce the vertical gradient of lung density and increase the amount of aerated lung.6 Consequently, in patients with ARDS, the application of PEEP appears to result in the opening of nonventilated or poorly ventilated alveoli.

In addition to serving as a purely mechanical stent, PEEP may keep alveoli patent by preserving surfactant function (which has been observed to be functionally altered in ARDS), and in so doing may reduce surface tension and reduce the tendency of alveoli to close.37,38 In a study evaluating the effect of PEEP, eight patients with ARDS initially underwent a lung recruitment procedure involving a period of sustained inflation.39 When PEEP was decreased below the inflection point, a decay in lung compliance occurred. Interestingly, this study found a decelerating waveform to be more beneficial in producing an improvement in compliance than a constant-flow waveform. It was hypothesized that the decelerating waveform produced an initial high airway pressure that was maintained throughout the respiratory cycle, allowing recruitment of alveoli.40 This concept is important and raises the possibility that peak airway pressures may in fact lead to alveolar recruitment, and that PEEP maintains the level of recruitment.

The relative importance of maintaining airway patency and using relatively high levels of PEEP is emphasized by recent observations that lung underdistention may be as injurious as lung overdistention and may contribute to the development of VILI. Ventilation without PEEP or at levels of PEEP that did not result in lung recruitment has been shown to cause respiratory and membranous bronchiolar injury, a reduction in lung compliance, and hyaline membrane formation.41,42 In theory, ventilation at low lung volumes causes repetitive opening and closure of alveoli. This in turn may lead to the development of shear stress along the bronchial and alveolar walls. Repetitive stress is known to disrupt surfactant and is speculated to disrupt epithelial structures and contribute to failure of the alveolar-capillary barrier.

PEEP also improves gas exchange and lung mechanics, potentially by redistributing lung water from the alveolar to the extra-alveolar interstitial space.43 Finally, PEEP also has significant hemodynamic effects and typically results in a reduction in ventricular preload and a reduction in cardiac output. Dreyfuss and Saumon postulated that the benefits of PEEP in ARDS stem from its effect on pulmonary perfusion, and demonstrated that the reduction of lung edema produced by PEEP was negated when dopamine was administered to keep arterial blood pressure constant.26 They concluded that it was end-inspiratory volume, and not end-expiratory volume (FRC), that was important in the development of VILI. A recent study by Duggan and coworkers has highlighted the importance of atelectasis.44 They evaluated the effects of a lung recruitment maneuver on lung and cardiac function. Animals whose lungs were not recruited (and presumably remained atelectatic) had a higher mortality, increased microvascular leak, worse oxygenation, and worse pulmonary vascular resistance. The increase in pulmonary vascular resistance may have led to the decrease in right ventricular function and increase in serum lactate in these animals.

In summary, in patients with ARDS, PEEP may improve lung compliance and oxygenation by recruiting alveoli and maintaining patency throughout the respiratory cycle (Fig. 37-3). Furthermore a ventilation strategy that fails to optimize end-expiratory volume with PEEP may contribute to VILI through the development of shear stress during repetitive opening and closing of lung units. Consequently, the notion of best PEEP needs to be extended to include an end-expiratory pressure level that will minimize lung injury.

Biotrauma

It is becoming increasingly clear that mechanical stretch or shear injury leads to inflammatory mediator release and cellular activation. Mechanotransduction—whereby a physical stimulus produces cellular signaling—has been an area of active investigation, and in the setting of mechanical ventilation this effect has been termed biotrauma.45,46 The notion that VILI may be responsible for the production of inflammatory mediators was highlighted in a study by Tremblay and associates in an isolated perfused rat model.47 They demonstrated that the use of an injurious ventilation strategy (zero PEEP, high tidal volume, or both) had a dramatic influence on lung inflammatory mediators. Since this study, mechanical ventilation has been shown to be capable of causing the regional production of a variety of proinflammatory mediators, including tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), interleukin-6 (IL-6), interleukin-10 (IL-10), macrophage inflammatory protein-2 (MIP-2), and interferon-γ (IFN-γ).48–55 Indeed both overdistention (volutrauma) and underdistention (atelectrauma) may be equally important in inducing inflammatory mediator release.51,56 However, as emphasized in an excellent review by Uhlig, it is important to differentiate the effects of cellular stretch– or shear stress–induced necrosis from true mechanotransduction, in which intracellular signaling occurs.46 Cellular necrosis would potentially lead to the release of preformed cellular inflammatory mediators, while stimulation of intact cells would potentially lead to the release of preformed mediators via alterations in the cellular cytoskeleton, and from stimulation of mRNA and production of new proteins. The fact that production and release of inflammatory mediators does occur was illustrated by Tremblay and colleagues in an isolated perfused rat model of VILI, whereby injurious patterns of mechanical ventilation led to an increase in production of c-fos mRNA.50

The concept of ventilator-induced mechanotransduction and subsequent inflammation is evolving and the cellular source of inflammatory mediators and subsequent lung injury is an area of active research.57,58 Both inflammatory cells and pulmonary epithelial cells appear to be capable of responding to mechanical stresses. Earlier workers attributed the increase in lung lavage cytokines predominantly to production by neutrophils and macrophages.59,60 More recent attention has focused on mesenchymal cells. Mechanical stretch of cultured fetal rat lung cells increased the expression of MIP-2, a potent neutrophil chemokine.58 Interestingly stretch alone did not lead to an increase in MIP-2 mRNA expression while stretch in addition to LPS did. This finding and the rapid increase in MIP-2 following stretch suggested that stretch alone leads to secretion and not synthesis of MIP-2. The same group later demonstrated increased expression of TNF-α and IL-6 mRNA in pulmonary epithelial cells following the application of an injurious ventilatory strategy, in an ex-vivo rat model.50 Importantly, they demonstrated that the epithelial cells were a major source of TNF-α. These findings are extremely relevant as they support the notion that mechanical stretching or injury to the lung epithelium is sufficient to stimulate an inflammatory response and potentially recruit neutrophils to the lung.59

In addition to epithelial cells, endothelial cellular stretch may also lead to the development of a proinflammatory state. Using a novel method of cellular recovery and evaluation of endothelial cells, Bhattacharya and coworkers demonstrated that lung stretch was capable of inducing P-selectin expression on the cell surface through a tyrosine phosphorylation signaling pathway.61 Indeed the importance of endothelial expression of cell surface receptors may be critical to the development of neutrophil accumulation in the lungs.62 Bhattacharya and coworkers also observed that lung stretch lead to the development of focal adhesions, suggesting that endothelial cells actively respond to the adverse mechanical milieu, and that the matrix proteins participated in this response. Several investigations have suggested that stretch-induced release of matrix proteins is an important mediator of VILI.63,64 In particular, the release of matrix metalloproteinases (a group of enzymes that are capable of damaging the extracellular matrix and stimulating fibroblasts) and fibronectin have been reported to be upregulated in models of VILI.58,65

The inflammatory effects induced by injurious ventilatory strategies may be attenuated by pretreatment with corticosteroids or heat stress (presumably by induction of heat shock proteins).48,66,67 Ohta and colleagues demonstrated that 250 mg/kg of methylprednisone suppressed markers of neutrophil activation in rats subjected to high-tidal-volume ventilation.66 Similarly Held and coworkers48 demonstrated that administration of dexamethasone blocked the release of nuclear factor-κB (NF-κB), cytokines, and chemokines in an isolated perfused mouse model of endotoxin-mediated acute lung injury and VILI. From a clinical perspective these observations are enticing. They also raise the possibility that the late phase of ARDS, characterized by fever and persistent airspace disease, in some patients may be mediated by ongoing ventilator-induced lung injury.68–70 Thus the experimental observation that corticosteroids are capable of modifying the inflammatory response from VILI may also account for the observed improvement in these patients with persistent ARDS following the administration of glucocorticoids.71

In summary mechanotransduction due to lung distention during mechanical ventilation appears to be capable of stimulating signaling pathways with secretion of inflammatory mediators in a variety of cells including neutrophils, macrophages, epithelial and endothelial cells, and the extracellular matrix. However, in the face of these observations, the notion that VILI contributes to mortality in patients with ARDS presents somewhat of a paradox, as most patients with ARDS do not typically succumb to progressive hypoxic respiratory failure from worsening lung injury. Rather, death in these patients is caused by the development of multiple organ dysfunction.72,73 However, recent experimental and clinical studies suggest that ongoing VILI is capable of causing nonpulmonary organ injury through the translocation of regional inflammatory mediators and bacteria into the systemic circulation.

Decompartmentalization

Damage to the alveolar-capillary barrier from mechanical ventilation may be central to several important pathophysiologic mechanisms in the development of VILI. First, loss of membrane integrity is key to the development of the pulmonary edema and hyaline membrane formation—the hallmarks of ARDS. The loss of barrier function may also allow for entry of inflammatory cells into the lung, in turn promoting the perturbation of lung structure and function. More recently, however, the loss of the integrity of the alveolar-capillary membrane has been proposed to be central to the spread of lung inflammation (partially produced by mechanical injury) to nonpulmonary organs, and thus may lead to their subsequent dysfunction.

The development of increased edema fluid during mechanical ventilation is in part due to an increase in filtration of fluid into the interstitial space.74,75 With progressive lung inflation, a decrease in interstitial pressure occurs.76 In addition, lung inflation causes dilation of extra-alveolar vessels. The net effect of progressive lung inflation, therefore, is an increase in the hydrostatic gradient for fluid to move from the capillaries into the interstitial space.76–78 In addition to an increase in filtration, ventilation with high lung volumes may also cause an increase in permeability of the alveolar-capillary barrier to proteins. Several studies have demonstrated an increase in filtration coefficient during mechanical ventilation.29,79 This effect is thought to be due to a reduction in the integrity of the alveolar-capillary barrier. During mechanical ventilation, injury has been demonstrated to occur to both the epithelial and endothelial membranes. Both edema and injury to the alveolar-capillary barrier can develop rapidly, with transient alterations in filtration coefficient occurring in animals ventilated for only 2 minutes.80 Subsequent studies have demonstrated that ventilation for more prolonged periods is associated with progressively worse edema, which becomes irreversible. In rats ventilated for 5 to 20 minutes, extravascular fluid accumulation increased with time of ventilation, progressing from perivascular edema to interstitial edema, and eventually to alveolar edema. Damage to type I cells, denudation of the basement membrane, and hyaline membrane formation were seen at 20 minutes. Thus mechanical ventilation appears to have a cumulative effect, with increased exposure to high airway pressures resulting in progressive and eventually sustained lung injury.

It seems logical that the movement of proteins would not be unidirectional from circulation to lung. Indeed recent evidence suggests that translocation of proteins occurs from lung to circulation as well. Using a rat model of VILI, Lesur and associates demonstrated an increase in total protein in lung lavage fluid with lung overdistention.81 Furthermore they demonstrated that Clara cell protein (found in high concentrations in Clara cells located in terminal bronchioles) decreased in lung lavage fluid with time, but increased in the serum of the rats. These observations are highly relevant and demonstrate that mechanical ventilation can lead to bidirectional translocation of proteins. Promotion of translocation of endotoxin and bacteria from the lungs to the circulation has also been demonstrated to be a feature of VILI.82–84 This loss of compartmentalization and resultant translocation of proteins, potential inflammatory mediators, and bacteria may be central to the development of more distal organ damage. The concern then is that the lung may become the machinery for multiple-organ dysfunction syndrome (MODS). Indeed Choi and associates recently demonstrated that large tidal volumes produced both pulmonary and renal microvascular leakage of protein and Evans blue dye.85 Interestingly, endothelial nitric oxide synthase (eNOS) expression was significantly increased in the affected lung and kidney samples. Their observation that inhibition of eNOS attenuated the degree of microvascular leakage supported the notion that eNOS contributed to the loss of the integrity of the membrane barrier. These observations were extended in a recent study by Imai and associates.86 They demonstrated that serum from rabbits that were ventilated with an injurious ventilation strategy was capable of increasing the rate of apoptosis in cultured epithelial cells from the kidney and villi of the small intestine (Fig. 37-4). Functionally this correlated with biochemical evidence of worsened renal function when intact animals were studied.

That distal organ injury in humans may be mediated by mechanical ventilation is supported by two observations. First, studies in humans have shown that an injurious ventilator strategy leads to an increase in circulating cytokine levels. Ranieri and coworkers randomized 44 patients to receive either mechanical ventilation to maintain normal blood gases (n = 19), or a lung-protective strategy attentive to both lung distention and maintenance of an adequate end-expiratory lung volume (above the lower inflection point; see below).53 Despite similar levels of circulating cytokines at baseline, patients in the control group had an increase in plasma concentrations of TNF-α, IL-6, and TNF-α receptors over a period of 36 hours. Conversely, patients in the lung-protective strategy group had a reduction in plasma concentrations of IL-6, soluble TNF-α receptor 75, and IL-1 receptor antagonist. Additionally they found that the concentration of the inflammatory mediators 36 hours after randomization was significantly lower in the lung-protective strategy group than in the control group. In a subsequent analysis of the serum in 11 of the controls and 9 of the lung-protective strategy group, levels of soluble Fas ligand (shown to induce apoptosis in human epithelial cells in vitro87,88) and creatinine were found to be elevated in the controls.89

Traditionally, the goals of mechanical ventilation have been to maintain adequate tissue oxygenation (generally accepted as a saturation greater than 90%, with a reasonable cardiac output) and adequate ventilation to achieve normocarbia and maintain normal blood pH. To achieve these goals, patients with acute respiratory failure were often ventilated using a standard ventilation cocktail consisting of variable levels of PEEP and oxygen, and a tidal volume in the range of 10 to 15 mL/kg. This strategy was often employed with little regard for the underlying cause of respiratory failure. However, based on the forgoing discussion, adopting such a strategy may be deleterious. Consequently clinical trials and experimental endeavors have focused on the development of lung-protective strategies. The principal objectives of a lung-protective strategy are to limit alveolar distention and maintain alveolar patency. A lung-protective strategy should be implemented early in the course of respiratory failure to prevent the development of VILI.

There are four potential strategies, none of which are mutually exclusive, to reduce alveolar overdistention. The first strategy is to limit inflation pressure. Current opinion and recommendations from a consensus conference on mechanical ventilation in ARDS recommend that plateau pressure be kept below 35 cm H2O in patients with ARDS. However, as emphasized in a review by Brower,90 a strategy that focuses on plateau pressure alone may still produce potentially injurious tidal volumes. Conversely, another limitation of using airway pressure as an indicator of lung inflation is that this method assumes that chest wall (including abdominal) compliance is normal. Patients with ARDS and conditions such as obesity or ascites may have a significantly reduced abdominal compliance. Consequently, high airway pressures may not be a reflection of overdistention. A second strategy to prevent lung overdistention is to limit tidal volume. This strategy is attractive because it is simple and eliminates concerns about changes in thoracic or abdominal compliance. In addition, this method was shown to be effective in decreasing mortality compared to conventional lung volumes.2 The potential difficulty with limiting volume is that the tidal volume used is still rather arbitrary and does not take into account the severity of lung injury. In fact, the ARDS Network study2 used a protocol that limited tidal volume to 6 mL/kg predicted body weight (PBW) and limited plateau pressures to less than 30 cm H2O. It is clear based on this study that tidal volumes of 12 mL/kg are excessive. What is not known at present is whether 6 mL/kg PBW volumes are superior to larger volumes (8 to 10 mL/kg). Whether 6 mL/kg is superior or inferior to 8 mL/kg remains a matter of debate. However, until evidence to support the contrary emerges, 6 mL/kg PBW should be considered the standard of care. It is worth emphasizing that the ARDS Network trial based tidal volumes on the patient's PBW, and not measured body weight. In this study a tidal volume of 6 mL/kg based on predicted body weight was on average equivalent to a tidal volume of 5 mL/kg measured body weight.

The Pressure-Volume Curve in Acute Respiratory Distress Syndrome

Some argue that ventilator settings should be individualized based on the pressure-volume (P-V) curve (see Fig. 37-3).24 Examination of the P-V relationship of the lungs of patients with ARDS provides a good conceptualization of the changes that occur during ARDS, and has been instrumental in the development of strategies of mechanical ventilation to reduce lung injury in these patients. Indeed one approach to optimize ventilation in patients with ARDS makes use of the shape of the P-V curve, in order to maintain lung inflation between the upper and lower inflection points. With a progressive increase in static airway pressure, there is a gradual increase in lung volume, which depends on the compliance of the respiratory system. As lung inflation approaches total lung capacity, a similar increase in airway pressure produces less lung inflation, and the P-V curve begins to flatten. This is often referred to as the upper inflection point. Continued inflation beyond this point results in overdistention of alveoli and increases the risk for macroscopic and microscopic lung injury. Similarly (although more controversial a notion) the lower inflection point is felt to reflect opening of atelectatic regions of the lung. It has been suggested that the level of PEEP be maintained above this level to avoid cyclic injury (atelectrauma) to the lung. However, this attractive approach has a number of problems. First, use of PEEP a few cm H2O above the lower inflection curve does not ensure that the lung is recruited; indeed recruitment takes place over the entire steep portion of the P-V curve.91–93 Secondly, the upper inflection point may indicate the completion of recruitment, rather than the development of overdistension.93 Furthermore, for the practicing intensivist, determining the lower inflection point may not be practical. Some guidance is provided by studies that suggest that in most patients with ARDS, PEEP levels above 15 cm H2O are generally required to maintain lung inflation above the lower inflection point.24,94

Permissive Hypercapnia

Permissive hypercapnia was first described in patients with asthma who were being mechanically ventilated. In an attempt to reduce dynamic hyperinflation and gas trapping, minute ventilation was decreased to allow sufficient expiratory time for the lung to empty via the obstructed airways. Relative to historical controls, patients managed in this manner had much better outcomes, with less barotrauma and less time spent receiving mechanical ventilation.95,96 The concept of permissive hypercapnia was extended to the management of patients with ARDS, and follows occasionally from a mechanical ventilation strategy that demands a restriction in tidal volume. It should be emphasized that permissive hypercapnia does not represent a method of mechanical ventilation per se. Rather, it is the consequence of a strategy that limits lung volume excursions in an attempt to minimize alveolar overdistention and hence VILI. In an uncontrolled study, Hickling and associates described the use of a pressure-limited strategy and permissive hypercapnia in patients with advanced ARDS. Mortality was significantly lower than that which would have been predicted from the APACHE II score alone.97 This was subsequently confirmed by some of the same authors in a prospective uncontrolled study in patients with ARDS.98 These were landmark studies, but used historical controls, the method of ventilation was not well defined, and efforts were made concurrently to limit oxygen toxicity. Therefore, these studies suggested, but did not conclusively prove, that a pressure-limited strategy and disregard for the partial arterial pressure of carbon dioxide (PaCO2) improved outcome. Importantly, serious side effects of an elevation of PaCO2 were not observed in either study.

The physiologic consequences of hypercapnia and respiratory acidosis have been reviewed extensively.99,100 At present, the only absolute contraindication to a rise in PaCO2 is increased intracranial pressure, although acute hypercarbia may have adverse effects on the fetus in gravid individuals. In patients with critical illness and impaired oxygen delivery, there have been concerns about effects on cardiovascular performance. However, the potential myocardial depressant effects are usually short-lived owing to the buffering capacity of myocytes and the increase in sympathetic activity and decrease in afterload that accompany hypercapnia.101 Caution, however, is warranted in patients with evidence of myocardial dysfunction. A final concern surrounds the increase in pulmonary arterial pressures that develops with hypercapnic acidosis. This response is likely mediated by a reduction in nitric oxide and has been shown to be reversible with nitric oxide inhalation.102 In the face of these concerns hypercapnia has been shown to be protective in a variety of settings.103–106 Indeed it has been postulated that hypercapnia may attenuate the severity of acute lung injury, or at the very least that the development of hypocapnia may be harmful.100,103,107,108 Hypercapnic acidosis has been shown to attenuate protein leakage, lung edema, lung lavage inflammatory mediators, and lung injury score, and preserve oxygenation and lung compliance in several models of lung injury (Fig. 37-5).1,108–110

More controversial is whether the resulting respiratory acidosis should be corrected by the infusion of a buffer solution. Although some advocate the use of a bicarbonate infusion, in the setting of impaired ventilation the additional CO2 generated during the buffering of H+ with bicarbonate might worsen intracellular pH. Second, as pointed out by Feihl and Perret, very large doses of bicarbonate are required to produce a significant improvement in pH during hypercapnia, owing to the large volume of distribution of the bicarbonate ion, to renal bicarbonate losses, and to conversion of bicarbonate to CO2.99 In addition, although extracellular pH is low, intracellular pH is rapidly corrected and seldom becomes critical. Consequently, the slow bicarbonate infusion recommended by some studies is not likely to be of benefit, and the routine administration of bicarbonate during permissive hypercapnia remains controversial. Indeed Laffey and coworkers demonstrated that correction of hypercapnic acidosis worsened lung function in an ischemia-reperfusion lung injury model.107 Other buffers such as Carbicarb may theoretically be more efficacious in the setting of hypercapnia.99,111

Clinical Trials

Based on the foregoing discussions several controlled clinical trials have evaluated the effects of a lung protective strategy on outcome in ARDS (Table 37-3). However, initial randomized clinical trials evaluating the effect of lower tidal volumes on outcome were disappointing.112–114 There was even a suggestion that tidal volume restriction was harmful, as it was associated with a greater use of neuromuscular blockers, a greater need for dialysis (perhaps related to the lower pH from a higher PaCO2), and a trend toward higher mortality. In the study by Stewart and associates, the mortality in the tidal volume restriction arm was 50% compared to the control arm mortality of 47%, while in the study by Brochard and colleagues, the mortality was 47% and 39%, respectively.112,114 However, the NIH-sponsored multicenter study of patients with ARDS has vindicated many of the earlier animal studies and clinical trials.2 In this trial patients were randomized to receive either “conventional” tidal volumes (12 mL/kg PBW; tidal volume was reduced if plateau pressure was greater than 50 cm H2O), or a lower tidal volume (6 mL/kg PBW, and maintenance of a plateau pressure between 25 and 30 cm H2O). The trial was stopped early after an interim analysis demonstrated a survival benefit in the group with low tidal volume (Fig. 37-6). Mortality was reduced by 22% from 40% in the conventional arm to 31% in the low-lung-volume arm (CI 2.4 to 15.3 percent difference between the groups). The benefit of a lung-protection strategy seemed to be independent of the severity of the lung compliance at baseline. In addition to a mortality effect, the number of days alive and free of mechanical ventilation was lower in the intervention arm. However, this effect was solely due to the reduction in mortality, as the median duration of mechanical ventilation was 8 days for survivors in both groups. The benefit did not appear to differ when patients were stratified based on their risk factor for ARDS.115 Interestingly the number of days with nonpulmonary organ failure was lower in the intervention arm, and the plasma interleukin-6 concentration was decreased compared to the control group. This again supported the notion that a lung protection strategy achieved its benefit through a reduction in the systemic release of inflammatory mediators and reduction in severity of multiple system organ failure. Unlike previous studies, however, there was no difference in the use of neuromuscular blockers.

It is difficult to reconcile the difference in the results of the ARDSNet study with earlier clinical trials evaluating a lung volume restriction strategy, because the ARDSNet study differed in several ways, making direct comparisons difficult.116 First, the method of determining predicted body weight (and hence tidal volume) was different from earlier trials. Second, patients in the low-tidal-volume arm had higher respiratory rates that may have led to significant auto-PEEP, in turn leading to improved alveolar patency or recruitment. Third, the respiratory acidosis was more likely to be corrected with bicarbonate. This may have reduced the number of patients dialyzed, and could have reduced some of the yet to be determined effects of hypercapnic acidosis.

A concern regarding the safety of the ventilation trials conducted in patients with ARDS has recently been raised. In a review of the controlled trials of mechanical ventilation in ARDS, Eichacker and associates presented the argument that 12 mL/kg was potentially excessive, and that the use of this tidal volume as the reference intervention was inappropriate, placing patients in the control arm at risk.117 The authors argued that there should have been a control group that better reflected “conventional” treatment. What tidal volume this control group would have actually been managed with is speculative. The reader is referred to an excellent review of the controversy and its consequences by Steinbrook.118 The implications for this issue for the design and conduct of subsequent trials remain to be seen, but at present the ARDSNet strategy for ventilation of ARDS patients should be viewed as the standard.

In addition to lung overdistention, VILI also incorporates the concept that underdistention of alveolar units can also lead to injury. At present there are only two clinical trials that have evaluated the effects of an “open lung” approach to patients with ARDS. The first study by Amato and colleagues examined the effect of a multifaceted strategy that (1) minimized tidal volume, (2) recruited alveoli through a sustained inflation, (3) used a level of PEEP above the closing pressure of the lung, and (4) utilized a pressure-volume curve to define the optimum lung volume and PEEP.119 Consequently the specific effects of maintaining alveolar patency cannot be determined from this trial. Nonetheless, using this strategy they demonstrated an impressive reduction in mortality. However, the major criticism of this study is that the control group was significantly disadvantaged by a protocol that allowed for significant overventilation, and that the observed results were not due to a benefit in the treatment arm, but rather a detrimental outcome in the control group.

In order to make specific recommendations about the optimum strategy, the relative effects of tidal volume reduction, alveolar recruitment, and level of PEEP need to be determined. To this end two multicenter trials are evaluating the effect of lung recruitment and high PEEP on outcome in ARDS. The ARDS Network has reported the results of their trial (ALVEOLI trial) that was closed after enrolling 550 patients, citing a finding of futility in the trial being able to determine a treatment effect during an interim analysis.124 In an ongoing Canadian trial over 800 patients will be randomized to either low tidal volumes (identical to the ARDSNet low stretch trial) or a low tidal volume with an open lung approach. The open lung approach will utilize both periodic sustained inflations to 40 cm H2O continuous positive airway pressure for 40 seconds and high levels of PEEP (up to 20 cm H2O). It is hoped that this strategy will ensure both alveolar recruitment and maintenance of alveolar patency, respectively. Preliminary evidence suggests that this strategy may lead to a significant improvement in oxygenation in a subset of patients with early ARDS and no impairment in chest wall mechanics.120 Whether this physiologic improvement and attendant reduction in atelectasis will lead to an improvement in survival has yet to be proven.

High-frequency oscillation (HFO) may accomplish many of the goals of a lung-protective strategy. It utilizes small tidal volume excursions at a high mean airway pressure. Consequently lung overdistention may be prevented and alveolar patency may be maintained. In a study of HFO in 70 pediatric patients with ARDS, Arnold and associates reported an improvement in oxygenation and requirement for supplemental oxygen at 30 days in the HFO group.121 However there was no difference in duration of mechanical ventilation or survival. These observed improvements in physiologic parameters were also found in a study by Derdak and coworkers.122 In a multicenter randomized controlled trial of HFO compared to what was at that time a conventional ventilatory strategy, in 148 adults with ARDS they found no difference in survival or duration of mechanical ventilation. However the oxygenation was better in the HFO group during the study. These reports suggest that HFO is at least as safe as conventional mechanical ventilation, and may be associated with a more rapid improvement in oxygenation. However, both trials were hampered by the fact that the control group likely did not represent the current standard of care, namely tidal volumes of 6 mL/kg PBW. It remains to be seen if HFO is any more efficacious than a strategy that restricts tidal volume (with or without a strategy to maintain alveolar patency).

Even in the face of overwhelming experimental evidence and convincing results of a large controlled study, one of the major challenges faced by intensivists seems to be implementing these recommendations in practice. Two studies have demonstrated that compliance with the ARDS Network tidal volume goals are poor, even in those centers that participated in the original trial. Clearly education and protocolization of care will need to occur to allow best evidence to be translated into clinical practice.3,116

ARDS continues to be a common component of multisystem organ dysfunction and primary lung injury. Unfortunately, no pharmacologic intervention has proven to be efficacious in reducing mortality in patients with ARDS. An improved understanding of ARDS and the notion that prior ventilator strategies may have been injurious has led to rethinking how these patients should be supported.

In addition to theoretical concerns and experimental data in animals, evidence is evolving that our traditional goals of ventilation need to be modified to incorporate a lung-protective strategy (see Table 37-1). The low-lung-volume ARDSNet trial demonstrated a reduction in mortality with the implementation of a lung-protective strategy. However, it is a matter of perspective as to whether the intervention produced benefit, or simply that harm was avoided. Given the complex nature of ARDS, it is likely that a multifaceted strategy that incorporates several principles of VILI will need to be adopted. Limiting tidal volume may only be a component of a lung-protective strategy. Support is evolving for attempting to recruit alveoli and for the use of liberal PEEP. Future research on the effects of mechanical ventilation on regional and distal cellular signaling, apoptosis, and distal organ injury need to be explored further, incorporating recent advances in genomic and proteomic methods.123