Beyond volutrauma in ARDS: the critical role of lung tissue deformation

Ventilator-induced lung injury (VILI) consists of tissue damage and a biological response resulting from the application of inappropriate mechanical forces to the lung parenchyma. The current paradigm attributes VILI to overstretching due to very high-volume ventilation (volutrauma) and cyclic changes in aeration due to very low-volume ventilation (atelectrauma); however, this model cannot explain some research findings. In the present review, we discuss the relevance of cyclic deformation of lung tissue as the main determinant of VILI. Parenchymal stability resulting from the interplay of respiratory parameters such as tidal volume, positive end-expiratory pressure or respiratory rate can explain the results of different clinical trials and experimental studies that do not fit with the classic volutrauma/atelectrauma model. Focusing on tissue deformation could lead to new bedside monitoring and ventilatory strategies.


Introduction
Mechanical ventilation is the cornerstone of treatment for acute lung injury (ALI) and the acute respiratory distress syndrome (ARDS). Like any other treatment, mechanical ventilation can also have adverse eff ects. Among these treatments, ventilator-induced lung injury (VILI) is one of the most studied. Since the fi rst descriptions of the physical factors (pressure, volume, and so forth) involved in VILI, many mechanisms that promote lung injury -from lung mechanics to biochemical responses -have been identifi ed [1], and ventilatory strategies aimed at reducing lung injury have been devised. Th e evidence for a clinical counterpart of VILI, which has been termed ventilator-associated lung injury, is more subtle, because the ventilatory settings are not as aggressive and the nature of lung injury in this case is always multifactorial [2]. Diverse clinical trials have studied the eff ects of protective ventilation strategies aimed at reducing ventilator-associated lung injury [3][4][5][6][7][8][9][10][11][12][13], but few studies have shown survival benefi ts [4,8,10].
In the current paradigm of VILI, high positive endexpiratory pressure (PEEP) should be one of the most powerful strategies to decrease VILI [14]. Recently published trials have failed to show a clear benefi t for high PEEP levels in a general population of patients with ALI and ARDS [9,11,12]. Th e mechanisms supposedly involved in VILI should therefore be reappraised to explain the failure of high PEEP levels and to design new therapeutic approaches to ALI/ARDS. Th e present article reviews the current paradigm of VILI and its limitations, and discusses additional explanations that can serve as a basis for the development of new strategies.

The classic paradigm of ventilator-induced lung injury
Th e classic paradigm of VILI maintains that there are two main physical triggers of lung injury during ventilation: overdistension of alveolar units (volutrauma) and cyclic changes in nonaerated lung (atelectrauma). High inspiratory lung volumes or pressures can cause injury through alveolar overdistension. Although there has been some debate about the primary force that causes injury, both volume and pressure are two sides of the same cointranspulmonary pressure. At the cellular level, stretching the lung beyond its capacity ruptures alveolar cell membranes [15], and the resulting cell death induces infl ammation. Moreover, subtler injuries to the cytoskele ton or extracellular matrix trigger infl ammation through intracellular signals [16].
Th e second mechanism of injury, cyclic changes in nonaerated lung, was deduced from the observation of lung injury during ventilation with low end-expiratory lung volume (EELV) or in the absence of PEEP. In this case, the mechanisms at the cellular level are less clear. Air bubbles fl owing through a collapsed or fl uid-fi lled airway might induce damage to the epithelium by

Abstract
Ventilator-induced lung injury (VILI) consists of tissue damage and a biological response resulting from the application of inappropriate mechanical forces to the lung parenchyma. The current paradigm attributes VILI to overstretching due to very high-volume ventilation (volutrauma) and cyclic changes in aeration due to very low-volume ventilation (atelectrauma); however, this model cannot explain some research fi ndings. In the present review, we discuss the relevance of cyclic deformation of lung tissue as the main determinant of VILI. Parenchymal stability resulting from the interplay of respiratory parameters such as tidal volume, positive end-expiratory pressure or respiratory rate can explain the results of diff erent clinical trials and experimental studies that do not fi t with the classic volutrauma/ atelectrauma model. Focusing on tissue deformation could lead to new bedside monitoring and ventilatory strategies.
generating a steep pressure gradient near the air bubble front [17]. A second consequence of low EELV can be a heterogeneous lung, and Mead and colleagues showed that the forces exerted on alveolar walls or septa in the interfaces between collapsed and aerated lung tissues can be amplifi ed, leading to cell injury [18].

Discrepancies with the model
Some observations from experimental and clinical research do not fi t with the classical paradigm. Th e application of PEEP is almost invariably associated with a decrease in VILI in diff erent experimental models of lung injury, such as high tidal volume (Vt) ventilation [14] or surfactant depletion [19]. When translating these models into clinical practice, a reduction in mortality has been found primarily in studies in which high PEEP is part of a greater strategy that includes low Vt [4,10]. Experimental designs combining high PEEP with low Vt make it impossible to clarify the contributions of each factor to the outcome. Th ese clinical results are far from the sometimes impressive eff ects of PEEP seen in experimental models.
Th e eff ects of PEEP in healthy lungs also result in discrepancies with the paradigm. In intact lungs, one would expect PEEP to produce overdistension and a lack of signifi cant cyclic recruitment; however, even in this setting, PEEP reduces VILI. One study showed that adding PEEP in intact rats ventilated with very high Vt could reduce injury [14]. Th ese fi ndings have a clinical correlate: a recent randomized trial showed in patients without lung injury that ventilation with PEEP did not worsen outcome, but improved oxygenation and decreased the risk of ventilator-associated pneumonia [20].
Th e eff ect of Vt on healthy lungs is also controversial. In experimental settings, a Vt of 10 to 15 ml/kg results in pressures of around 10 to 20 cmH 2 O and can induce substantial damage in a short time [21]. A classic experiment from Mascheroni and colleagues demonstrated that, even in spontaneously breathing animals, chemically-induced hyperventilation could trigger substantial lung damage [22]. Th is work highlights the importance of tissue deformation, represented by increased transpulmonary pressures, even during negative-pressure ventilation (absence of high alveolar pressure, more homogeneous infl ation). Again, the signifi cance of these fi ndings in clinical practice is not so clear. Studies that have addressed this issue in surgical patients have found no diff erences in the lung infl ammatory response or outcome between low Vt or high Vt with short ventilatory times (from 1 to 3 hours) [23,24], but found that high Vt increases proinfl ammatory mediators with longer venti latory times [25][26][27]. Moreover, observational data [28,29] and recent clinical trials [30,31] suggest that high Vt increases the risk of acute lung injury in critically ill patients. Th e absence of specifi c markers of ventilator-associated lung injury precludes insight into the mechanisms underlying these results, but data from these articles suggest that Vt is important even in the absence of relevant recruitment/ derecruitment processes (as these were patients with healthy lungs) and over distension (as pressures were below 30 cmH 2 O).
Low EELV has proved safe in both animals and patients, and this approach has sometimes been termed permissive atelectasis. In a recent study using an isolated lung model, Fanelli and colleagues demonstrated that permissive atelectasis caused the same amount of lung injury as the open lung strategy and only subtle diff erences in apoptosis and ultrastructural changes favoring the open lung strategy [32]. In the clinical setting, mortality rates in descriptive studies for patients managed with a low-pressure strategy are similar to those in clinical trials [33]. Taken together, these data contradict the volutrauma/atelectrauma model, suggesting that other mechanisms could be responsible for VILI.

Alveolar instability and ventilator-induced lung injury
Breathing, airfl ow, and blood fl ow exert complex mech anical forces on the lung that can be classifi ed in specifi c physical terms: stress is the force per unit area; strain, also called stretch or deformation, is the change in length in relation to the initial length; and shear stress is the force per unit surface area in the direction of fl ow exerted at the fl uid-surface interfaces. When applied to respiratory physiology, stress has been defi ned as transpulmonary pressure, and strain defi ned as the relationship between Vt and EELV. Although Gattinoni and colleagues have used the volume at zero PEEP as the reference EELV [34], it is not clear how recruitment (an increase in lung volume without increasing strain) can aff ect these measurements.
A series of mechanisms transform the lungs from a highly unstable structure (comparable with air bubbles in soap) into a very stable one. Th e surfactant system, alveolar interdependence, collateral ventilation, extracellular matrix, and mechanical properties of the chest wall work together to avoid alveolar collapse. Th ese mecha nisms guarantee a minimal resting volume (functional residual capacity/EELV) at end-expiration and a large number of alveoli to distribute each Vt. Th e net result is that healthy lungs present minimal changes in their structure during ventilation and only minor variations in alveolar size and shape. Using in vivo microscopy, several authors demonstrated that the change in size of subpleural alveoli is negligible when the total lung capacity remains between 10 and 80% [35,36], although substantial heterogeneity remained during alveolar expansion in normal lungs. Whereas some zones undergo only minimal changes in each breath, other zones undergo deformations of up to 20% of their length. Th is fi nding could have important consequences for VILI in healthy lungs.
Failure of any of the above-mentioned mechanisms leads to alveolar instability, understood as an excessive tissue deformation during each breath. Although this may be a vague term, and in fact the thresholds of stress and strain that lead to tissue injury are unknown, we will use it as a concept beyond the classic opening and closing hypothesis. In this sense, alveolar instability refers not only to tidal changes in aeration, but also to those circumstances of excessive lung tissue deformation irrespective of the alveolar initial status. Th is common mechanism could help to override the diff erent hypo thesis on lung injury induced by cyclic changes in lung aeration (that is, opening and closing of alveoli versus fl uid and foam in airways and alveoli [37]). Most circumstances that increase alveolar instability also increase the risk of VILI. Whereas the current paradigm considers alveolar insta bility and VILI to be concomitant consequences of an unknown cause, we hypothesize that alveolar instability is the main mechanism in VILI. Lung injury increases alveolar instability, and alveolar damage starts in the unstable zones [19]; VILI could spread from these zones where tidal changes are more relevant. Th ese zones are also present in healthy lungs, so this model could explain the induction of VILI in patients with and without previous lung injury.
Some studies have explored how ventilatory parameters infl uence lung deformation (Figure 1). Halter and colleagues demonstrated that changes in subpleural alveoli correlate directly to Vt and inversely to PEEP [38]. Interestingly, even with high Vt (up to 15 ml/kg), increasing PEEP levels from 5 to 20 cmH 2 O decreased alveolar instability from 108% to 15%. Experiments performed in cells corroborate these fi ndings. Using cell cultures submitted to cyclic stretch, Tschumperlin and colleagues showed that the magnitude of deformation was more important than the peak stretch in inducing cell death [39].
Excessive tissue deformation can also explain why healthy lungs ventilated with high Vt and moderate pressures develop lung injury. Vt promotes a timedependent increase in alveolar instability that could lead to lung damage [40]. Retrospective clinical studies have found that high Vt increases the risk of lung injury [28,29], and a recent randomized clinical trial found a higher risk of ARDS with 10 ml/kg Vt than with 6 ml/kg Vt [30].
Considering alveolar instability as the main mechanism of VILI explains how PEEP may reduce VILI [38]. Even during ventilation with high Vt, PEEP can signifi cantly reduce tidal changes in alveolar size. Valenza and colleagues found that VILI was delayed in intact rats ventilated with high airway pressures and PEEP [14]. One could hypothesize that PEEP avoids the progressive impairment in alveolar stability, as seen during in vivo microscopy studies.

Measurements of lung stability at the bedside
Explained in terms of stress and strain, ventilation with moderate/high PEEP and low Vt will reduce both stress (Vt) and strain (higher EELV induced by higher PEEP), thus reducing the risk of superimposed ventilator-associated lung injury. Terragni and colleagues found that even when Vt and plateau pressure are limited, patients with larger nonaerated lung compartments measured with computed tomography are exposed to tidal hyperinfl ation, with excessive lung deformation of the aerated compartment at each tidal breath [41]. Similarly, in patients with focal ARDS ventilated with the ARDSnet protocol, using a physiologic approach to PEEP setting based on the shape of the airway pressure curve as a function of time during constant fl ow attenuated alveolar hyperinfl ation [42]. Lack of lung recruitment with PEEP during protective lung ventilation is again associated with excessive lung deformation.
Chiumello and colleagues found that the plateau pressure and Vt were inadequate surrogates for lung stress and strain due to considerable overlap at both diff erent Vt and PEEP values, and strain was clearly overestimated when recruitment was computed [34]. Th ese results accounted for the high variability in functional Diff erences between end-expiratory and end-inspiratory alveolar size increase with tidal volume and decrease with positive end-expiratory pressure (PEEP) (measured in an experimental model of lung injury). Note the synergistic relationship between these two parameters, and that PEEP may decrease the change in size in spite of high tidal volumes. Data extracted from [38].
residual capacity (a hallmark of ARDS) and the ratio of lung elastance to total respiratory system elastance. Experi mental models in which relatively low pressures may induce severe injury, discussed in previous sections, also reinforce this inaccuracy of airway pressures and volumes to estimate tissue injury, due to diff erences in specifi c lung and chest wall elastances. In this setting, only surrogates of stress and strain may help to minimize the damage induced by ventilation. When lung recruitability was taken into account, however, reducing the amount of opening and closing lung tissue by increasing PEEP yielded benefi ts that prevailed over the harmful eff ects of increasing alveolar strain [43]. Interestingly, opening and closing lung tissue was distributed mainly in the dependent and hilar lung regions and appeared to be an independent risk factor for death.

Reinterpreting the clinical trials
Th e ARDSnet trial on Vt demonstrated that overall survival is higher using a Vt of 6 ml/kg than using a Vt of 12 ml/kg at similar levels of PEEP; that is, similar lung volume at end-expiration [8]. We would expect much greater lung deformation in patients ventilated with higher Vt. Lowering Vt therefore not only reduces overdistension, but also improves alveolar stability.
Despite the controversy on how much high transpulmonary pressure a patient with ARDS can tolerate for how long and the sustained clinical benefi ts of this approach, very high plateau pressure in patients with ALI/ARDS and normal chest wall compliance can infl ict direct damage to the lungs or can aggravate existing disease. Nevertheless, recent trials have challenged the concept that mortality decreases directly with reductions in plateau pressure. Several studies tested the eff ect of two levels of PEEP on patients with ALI/ARDS ventilated with low Vt [9,[11][12][13]. At equal Vt, high PEEP resulted in higher plateau pressure at end-inspiration compared with low PEEP. Interestingly, despite the signifi cant diff erences in plateau pressures, no diff erences in mortality were observed, and only a trend toward lower mortality and fewer complications among ARDS patients was observed in the high PEEP group.
Th e benefi ts of high PEEP in ARDS patients were recently confi rmed in a large systematic review and metaanalysis [44]. Th e patients included in these studies were exposed to similar stress, whereas strain depended on the previous eff ect of PEEP in the lung parenchyma (recruitment or overdistension). Figure 2 illustrates the diff erential eff ects of ventilatory parameters according to the presence of recruitment. In ALI/ARDS, the percentage of potentially recruitable lung is extremely variable and is strongly associated with the response to PEEP. Th is may also explain why the benefi cial eff ects of PEEP are only seen in the most severe patients (that is, ARDS patients), in which recruitment and alveolar stabilization are the prevalent eff ects, whereas in those patients with mild injuries (ALI) this benefi t is lost in the face of increased overstretching [44]. Unfortunately, prediction of the potentially recruitable lung using physiological variables that can be measured at the bedside is neither specifi c nor sensitive [45], and PEEP induces both recruitment and overdistension in most patients [46,47].

Increasing the complexity of the model
Although the main focus of the present review is to explain how the classic concepts of volutrauma and atelectrauma can be unifi ed by understanding the eff ects of ventilatory parameters on lung structure and alveolar instability, mechanical ventilation can also have eff ects beyond the lung.
Intrathoracic pressures have a strong eff ect on hemodynamics, and Vt and PEEP can modulate right ventricular function, with consequences for lung injury [48]. Studies in experimental models of isolated ventilated and perfused heart-lung blocks have demonstrated that the lung hemodynamics may interact with tidal ventilation to modulate lung injury [49]. Infl ation of the lungs above their functional residual capacity results in an increase in pulmonary vascular resistance [50], and the increase in vascular resistance is largely the result of compression of the alveolar microcirculation by a tidal increase in alveolar pressure and thus in extramural pressure. Tidal alveolar vessel compression is opposed by the alveolar vessel intramural pressure, which, everything else being equal, varies in parallel with left atrial pressure. In other words, an increase in or a reduction in left atrial pressure, respectively, opposes or enhances the alveolar vessel compression from positive pressure ventilation [51]. Recent data have given credit to the concept that lung vessels exposed to injurious ventilatory patterns behave like material prone to fatigue and failure in similar experi mental conditions. It thus appears that both high and low capillary pressures are best avoided to limit the risk of VILI [49,51]. Moreover, these eff ects are aggra vated by ventilation at rapid respiratory rates with conco mitant increases of pulmonary artery pressure [52] and in conditions of high cardiac output [53,54]. In this context, PEEP serves to diminish transmural pressure (decreases strain) and attenuates edema formation and lung hemorrhage [48]. Th ese data highlight the inter active nature of the processes and cofactors in the modulation of VILI-like global and/or regional pulmo nary hemodynamics; the presence, type, and timing of secondary lung insults; and interactions between regional heterogeneity of pulmonary perfusion and regionally heterogeneous peak airspace strains. Lastly, these vascular variables also hint at potential diffi culties in designing appropriate trials of lung-protective ventilatory strategies. Furthermore, mechanical ventilation can infl uence both local and systemic immune responses [1] because mechanical forces within the lungs trigger diff erent biological responses that spread into the systemic circulation. Although much knowledge about this issue has been generated in recent years, the specifi c contribution of ventilator settings to the infl ammatory response (that is, the biological activity brought about increasing volutrauma or atelectrauma) is unknown, and this complex response is likely to be infl uenced by patients' baseline conditions.

Conclusions
Knowledge of the mechanisms involved in VILI is important to develop ventilatory strategies that could result in clinical benefi ts. Recent advances suggest that isolated mechanical forces (plateau pressure, PEEP) cannot adequately explain VILI; rather, the amount of damage depends on the simultaneous interaction of these forces and the previous status of the lung parenchyma. Ultimately, dynamic changes in alveolar structure could be responsible for lung injury. Th e translation of these concepts to the bedside requires complex physiological reasoning and directed research. Eff ects of recruitment in alveolar stability. Alveolar stability can be achieved by the equilibrium between end-inspiratory and endexpiratory volumes. This stability allows defi nition of a hypothetical safe zone (blue area). Because strain can be viewed as the ratio between these volumes, this zone corresponds to normal or lower strain values. On the contrary, excessive end-inspiratory deformation in lungs with a low resting volume leads to alveolar instability (red area). The arrows represent how the tidal volume (Vt) and positive end-expiratory pressure (PEEP) modify the end-inspiratory/end-expiratory ratio. In recruitable lungs, PEEP induces an increase in end-expiratory lung volume that allows ventilation to stay in the safe zone, and strain decreases. In the absence of recruitment, however, both the Vt and PEEP lead to a predominant increase in endinspiratory volume, therefore increasing strain. FRC, functional residual capacity.