Mechanical Ventilation in Acute Respiratory Distress Syndrome

Luciano Gattinoni, Eleonora Carlesso and Pietro Caironi

Since the first description of acute respiratory distress syndrome (ARDS) in 1967,¹ mechanical ventilation has been the primary “buying time” treatment for acute lung injury (ALI). Mechanical ventilation is not a “gas exchanger,” but instead replaces, totally or partially, the force normally generated by the respiratory muscles to promote ventilation, providing muscle rest. The effects of mechanical ventilation on gas exchange are indirect and include (1) better clearance of carbon dioxide (CO₂) by the power of the mechanical ventilator to expand the diseased and collapsed lung; (2) improvement in oxygenation by preventing alveolar hypoxia caused by hypoventilation; (3) improvement in oxygenation by increasing inspiratory oxygen fraction, which affects alveolar partial pressure of oxygen (PAO₂); (4) improvement in oxygenation by opening lung regions otherwise collapsed (alveolar recruitment); and (5) improving oxygenation by maintaining positive end-expiratory pressure (PEEP), preventing the collapse of the lung regions previously recruited during the inspiratory phase.

The negative effects of mechanical ventilation are interwoven with its positive effects. First, the driving force applied by the ventilator, if excessive, may alter, in different ways, the lung parenchyma and produce ventilator-induced lung injury (VILI). Second, the positive-pressure ventilation unavoidably turns the physiologically negative intrathoracic pressures into positive pressures. In consideration of this, the guiding principle of mechanical ventilation of ARDS patients should be that, whenever one manipulates the ventilator settings, one should refer to one polar star: The new setting is less harmful to the lung structure than the previous one.

History

Extensive reviews on the history of mechanical ventilation of ARDS can be found elsewhere.2,3 In the 1970s, mechanical ventilation was recommended and performed with low PEEP and high tidal volume: “. . . larger tidal volumes (10 to 15 per kilogram) are preferable, having been used in several thousand ventilated patients with no evidence of development of pulmonary damage.”⁴ Today, this advice seems inconceivable. At that time, the main concerns were the putative harm of high inspiratory oxygen fraction and the hemodynamic impairment. It was later recognized in experimental and clinical settings^5–7 that high-volume/high-pressure mechanical ventilation could severely damage the lung parenchyma. Such lesions, primarily attributed to the excessive airway pressure, were collectively classified as barotrauma. In the same period, Suter and colleagues⁸ published a report that, for the first time, systematically described the interaction between PEEP, lung mechanics, gas exchange, and hemodynamics.

In the 1980s, based on the work of Dreyfuss,9,10 the focus progressively shifted from the potential harm of pressure to the harm of volume (overdistention), a concept that was popularized as volutrauma.¹¹ In the mid-1980s, the application of computed tomography (CT)^12,13 and the quantitative approach to CT analysis¹⁴ led to the concept of baby lung,^15,16 which accounted for most of the previous observations on respiratory mechanics, gas exchange functionality, and potential harm of mechanical ventilation. The premise of this line of thought is that high pressure or excessive distention applied to a small fraction of the lung parenchyma (with a size similar to the lung of a baby) unavoidably leads to structural lesions of the lung regions open to ventilation. Total lung rest was achievable with the use of extracorporeal CO₂ removal, targeted to prevent the damages of high-pressure/high-volume ventilation.^17–19 The target of mechanical ventilation shifted toward lung protection, rather than normal gas exchange functionality. Hickling and associates²⁰ proposed the “permissive hypercapnia” strategy for ARDS, providing “gentle treatment” of the portion of the lung that remains open to ventilation (the baby lung) through less aggressive mechanical ventilation, at the acceptable price of an abnormally higher arterial partial pressure of carbon dioxide (PaCO₂).

A large amount of experimental and clinical data over the years has supported the approach of a lung protective strategy.21–25 Low tidal volume prevents the excessive global stress and strain of the baby lung, whereas higher PEEP prevents the regional excessive stress and strain by avoiding alveolar collapse and reopening during mechanical ventilation.²⁶ These mechanical events are associated with an inflammatory reaction of the epithelial and endothelial lung cells (biotrauma/atelectrauma) as shown by the seminal study of Tremblay and colleagues²⁷ (see also work of Dreyfuss and Saumon¹¹ and Tremblay and Slutsky²⁸). The literature to date supports the thought that the harm of mechanical ventilation is due to excessive global or regional stress and strain.²⁹ This situation leads to two results—a physical rupture of the lung and a mechanically induced inflammation of the lung parenchyma, constituting VILI. The best “mechanical ventilation” would provide adequate gas exchange with the lowest amount of VILI.

Physiologic Basis of Mechanical Ventilation

Driving Force

To ventilate the respiratory system, force is required. The driving force for ventilation under normal circumstances is provided by the respiratory muscles. During spontaneous breathing, the thoracic cage expands. This causes a decrease in intrathoracic pressure (pleural pressure, Ppl) relative to the atmosphere (in normal conditions approximately 2 mm Hg of ΔPpl is sufficient to expand the thoracic cage by 0.5 L). Because the lungs are connected in series with the thoracic cage, their volume is expanded to a near equal extent (not considering the blood shift ³⁰). The force that distends the lung is the pressure difference between the alveoli and the pleural cavity (the transpulmonary pressure, PL). As the lung expands, the alveolar pressure becomes subatmospheric, and gas flow is generated (inspiration). When the respiratory muscles relax, the potential energy accumulated in the respiratory system (lung and chest wall) returns the chest wall and the lung parenchyma to the resting position (expiration). In spontaneously breathing subjects, the driving force (muscular pressure, ΔPmusc) is spent partly to expand the chest wall (ΔPpl), partly to expand the lung (ΔPL), and partly to overcome the resistances to the gas flow. In quasi-static conditions, in which the resistances to gas flow are negligible:

During positive-pressure mechanical ventilation, the rules are exactly the same, with the exception that ΔPmusc is substituted by the force provided by the ventilator, and the force is applied to the lung parenchyma and not to the chest wall. The driving pressure, which coincides with the airway pressure at plateau during an end-inspiratory pause (in quasi-static conditions), is spent first to expand the lung and second to expand the chest wall:

In spontaneous breathing and during positive-pressure mechanical ventilation, the distending force of the lung is the transpulmonary pressure. During spontaneous breathing, pleural pressure is negative (facilitating venous return), whereas during mechanical ventilation it is positive (impairing venous return). What matters from the pulmonary perspective is the distending force (i.e., the transpulmonary pressure [ΔPL]), whereas what matters from the hemodynamic perspective is the pleural pressure (ΔPpl).

Transpulmonary Pressure

At the same driving force applied to the whole respiratory system (lung and chest wall), the resulting transpulmonary pressure, ΔPL, may be extremely variable. If the lung is relatively “stiff,” and the chest wall is relatively “soft” (e.g., during pulmonary fibrosis or ARDS of pulmonary origin), a greater fraction of the driving pressure is spent to distend the lung (high transpulmonary pressure). In contrast, if the lung is relatively soft, but the chest wall is relatively stiff (e.g., in the presence of an abdominal disease or severe obesity), most of the driving force is spent to move the chest wall (high pleural pressure).

To express this phenomenon quantitatively, it is convenient to consider the concept of elastance. The elastance of the whole respiratory system is the driving force (ΔPaw) required to increase the lung and the chest wall 1 L above their resting position (Etot = ΔPaw/1 L). Part of this driving force is spent to increase the lung volume of 1 L (EL = ΔPL/1L), and part is spent to increase the chest wall by the same amount (Ecw = ΔPpl/1 L). Transpulmonary pressure (ΔPL) can be expressed as the driving force times the ratio between lung elastance and total elastance of the respiratory system:

The transpulmonary pressure for a given driving pressure uniquely depends on the ratio of lung elastance to respiratory system elastance. In normal subjects EL/Etot is approximately 0.5, whereas in patients with ARDS it may range from 0.2 (e.g., in obese patients or in patients with high intra-abdominal pressure) to 0.8 (e.g., in patients with a very small baby lung and a normal chest wall elastance). This variability implies that for the same driving force applied and read on the ventilator display (e.g., 30 cm H₂O), the resulting transpulmonary pressure may range from 6 to 24 cm H₂O (Fig. 11.1).

Figure 11.1 Two examples of EL/Etot variability in patients with acute lung injury/acute respiratory distress syndrome. A, EL/Etot = 0.2. B, EL/Etot = 0.8. This variability implies that for the same airway pressure applied (e.g., 30 cm H₂O), the resulting transpulmonary pressure (PL) may range from 6 to 24 cm H₂O.

Force-Bearing Structure of Lung Parenchyma

The transpulmonary pressure is applied to the force-bearing structure of lung parenchyma, the extracellular matrix, which constitutes the lung skeleton.³¹ The lung skeleton is a complex and metabolically active structure that includes a network of several components—elastin, collagen, and proteoglycans. All these molecules are involved in determining the mechanical characteristics of the respiratory system. The elastin may be considered as an elastic spring, whereas the unextensible collagen, which is folded at end expiration and completely unfolded at a lung volume equal to total lung capacity, acts as a stop-length fiber.^32,33 The proteoglycans stabilize the collagen-elastin network, contributing to lung elasticity and alveolar stability at low and medium lung volumes.³⁴

The matrix of elastin, collagen, and proteoglycans is arranged in two main fiber systems: (1) the axial system, which originates from the pulmonary hilum and runs deeply into the lung parenchyma down to the alveolar level, where it joins (2) the peripheral system, which originates from the visceral pleura and runs centripetally within the lung parenchyma.³¹ The lung skeleton may be considered as a continuous elastic structure that reaches its extension limits at total lung capacity, a lung volume equal to about threefold the lung resting volume. At this level of alveolar distention, the collagen is completely unfolded, and further expansion is prevented. The epithelial and endothelial cells do not directly bear the applied forces because they are anchored to the extracellular matrix by a series of structural proteins (integrins), which are connected to the cytoskeleton. During lung expansion, the epithelial and the endothelial cells modify their shape.

It is well documented that mechanically induced cellular deformation activates a series of mechanosensors with the production of several inflammatory mediators, such as cytokines (interleukin 6, tumor necrosis factor-α, and interferon-γ),27,35 metalloproteinases (enzymes involved in the remodeling of the matrix),³⁶ leukotrienes,³⁷ and interleukin 8,^38–40 the most powerful attractor of neutrophils.⁴¹ Gross barotrauma (e.g., pneumothorax) is due to the stress at rupture of the lung skeleton, whereas intrapulmonary inflammation is primarily due to the excessive strain of the epithelial and endothelial cells.

Concept of Stress and Strain

Stress is the applied force, and strain is the linear deformation of material. In the whole lung, the rough approximation of stress is the transpulmonary pressure, whereas the approximation of the average strain is the change in volume relative to the lung resting volume. The ratio between alveolar stress and strain is defined as lung-specific elastance (Espec), which is mathematically defined as:

where ΔV is the volume variation applied to the lung (i.e., the tidal volume), and V₀ is the lung resting volume (i.e., the functional residual capacity at atmospheric pressure [without any application of PEEP]). The lung-specific elastance is the transpulmonary pressure required to double the lung resting volume (i.e., the ΔPL when ΔV/V₀ is equal to 1). In ARDS, lung-specific elastance is similar to normal,16,42 reinforcing the concept of the baby lung (lung is small and not stiff), and questions the use of normalizing the tidal volume to the ideal body weight. The same tidal volume per kilogram may result in completely different strain according to the size of the baby lung (the V₀ of the previous equation). For example, a 70-kg man with ARDS may have, according to the severity of the lung injury, a residual baby lung equal to 60%, 40%, or 20% of his normal lung size. If the ventilator is set to deliver 10 mL/kg, the actual delivered tidal volume would generate an alveolar strain, which would result from the application, in normal lung, of a tidal volume equal to 17 mL/kg, 25 mL/kg, and 50 mL/kg, values associated with a significant lung injury in laboratory studies.^11,29

Recently we attempted to quantify the relationship between stress-strain and VILI in healthy animals. We found that edema formation was a threshold phenomenon, induced by mechanical ventilation when the global strain reaches a critical value of about 2.⁴³ This threshold roughly corresponds to the point where the stress-strain curve loses its linearity and starts an exponential growth, indicating that some lung regions reach their own total capacity and cannot expand any further. At this level of strain, in a period of 24 to 48 hours the mechanical ventilation is lethal and the increased lung weight (two to three times the baseline) is associated with a striking impairment of respiratory mechanics, gas exchange, hemodynamics, inflammation, distal organs damage, and 100% mortality rate. This lethal strain, and associated stress, however, is rarely applied in clinical practice. To explain VILI in a diseased lung, therefore, alternative phenomena must be taken into account as the lung dishomogeneity and the presence of stress rise, which will be discussed later.

Patient Characterization

Using the least harmful mechanical ventilation setting requires a preliminary knowledge of the main pathophysiologic characteristics of the patient. The most relevant characteristics are the kind and amount of the lung injury, the chest wall elastance, and the lung recruitability.

Gas Exchange

Oxygen

PaO₂, PaO₂-to-fraction of inspired oxygen ratio (PaO₂/FIO₂), and Riley’s shunt fraction ⁴⁴ are the most commonly used variables to assess the severity of lung injury. In particular, the PaO₂/FIO₂ thresholds of 300 and 200 were used to define ALI (300) and ARDS (200) according to the American-European Consensus Conference on ARDS.⁴⁵ Consequently, the PaO₂/FIO₂ ratio is perceived as a key variable by most intensive care unit (ICU) physicians: The lower the PaO₂/FIO₂ ratio, the greater the lung injury. This equivalence is highly questionable. First, in most large studies on ARDS, no association was found between hypoxemia and outcome.⁴⁶ In the large study showing a better outcome with low tidal volume compared with higher tidal volume, the PaO₂/FIO₂ ratio was significantly lower in the low tidal volume group despite ending with better outcome.²⁵ Finally, the PaO₂/FIO₂ ratio was not different in patients with early, intermediate, and late ARDS, suggesting that oxygenation was not dependent on the structural changes of lung parenchyma occurring with time.⁴⁷ In a study⁴⁸ in which the lung severity was assessed by CT scan (and defined as a fraction of the gasless tissue), we did not find significant changes of PaO₂/FIO₂ ratio over a wide range of nonaerated tissue (Fig. 11.2), and PaO₂/FIO₂ ratio was not associated with outcome.

Figure 11.2 Frequency distribution of nonaerated lung tissue, expressed as a proportion of the total lung weight, recorded at 5 cm H₂O positive end-expiratory pressure in 68 patients with acute lung injury/acute respiratory distress syndrome (ALI/ARDS). Green columns represent patients with PaO₂/FIO₂ less than 300 (ALI without ARDS), whereas purple columns represent patients with PaO₂/FIO₂ less than 200 (ARDS). The nonaerated lung tissue was defined as the lung tissue having a physical density at computed tomography scan image analysis between +100 Hounsfield units (HU) and –100 HU, representing the portion of lung parenchyma that is consolidated or collapsed or both (i.e., the lung injury severity). (With permission from Gattinoni L, Caironi P, Cressoni M, et al: Lung recruitment in patients with the acute respiratory distress syndrome. N Engl J Med 2006;354:1775-1786.)

Most data suggest that PaO₂/FIO₂ ratio is a weak indicator of the overall lung severity, with compensatory rearrangement of perfusion during ARDS limiting the deterioration of oxygenation. The same limits apply when PaO₂/FIO₂ ratio changes are used to assess lung recruitment. Because this maneuver is unavoidably associated with changes of perfusion (global or regional), the increase of PaO₂/FIO₂ ratio may be partly due to decrease of perfusion, as shown in the 1980s.49–51 Most data suggest that the use of oxygenation variables alone to assess lung severity is misleading.

Carbon Dioxide

Although less considered, the variables derived from CO₂, as the total or alveolar deadspace, seem to be of greater value in assessing lung severity. It has been shown in ALI/ARDS patients that deadspace at entry is a strong predictor of outcome,⁵² and that PaCO₂ for the same total ventilation steadily increases from early to intermediate and to late ARDS, reflecting the lung structural changes.⁴⁷ The PCO₂ response to prone position (in contrast to PO₂ response) is a strong prognostic index of mortality.⁵³ Most data suggest that CO₂-related variables (deadspace), more than PaO₂/FIO₂ ratio, reflect the severity of lung injury (and associated mortality rate) at the time of presentation and the structural changes of lung parenchyma occurring with time (fibrosis, Pneumocystis, and possibly perfusion defects).

Respiratory Mechanics, Chest Wall Elastance, and Lung Volume

In the original description of ARDS,¹ the low compliance (i.e., high elastance) of the respiratory system was a landmark of the syndrome. The respiratory system compliance is not considered in the current definition of ARDS, however.⁴⁵ For years, the low compliance was attributed uniquely to the lung component (lung stiffness and lack of surfactant). Quantitative CT shows, however, that the respiratory system compliance primarily reflects the size of lung open to gases (the baby lung^14,16), suggesting that the intrinsic functioning lung elasticity in ARDS is close to normal (the lung is “small” rather than “stiff”). More than gas exchange, the respiratory system compliance indirectly assesses the lung injury severity (the smaller baby lung, the greater lung injury),⁵⁴ as confirmed in animal experiments.⁵⁵ Another variable also must be taken into account—the elastance of the chest wall. It has been shown in a significant portion of ALI/ARDS patients that the chest wall elastance is greater than normal because of increased intra-abdominal pressure, obesity, or severe edema.⁵⁶ In patients with extrapulmonary ARDS⁵⁷ and in obese patients,⁵⁸ the high elastance of the respiratory system may be due to the chest wall and to the lung derangement. Measurement of intra-abdominal pressure should be considered when selecting mechanical ventilator settings.

Severity of Lung Injury and Lung Recruitability

CT can be used to assess the severity of lung injury by measuring the fraction of nonaerated lung tissue at end expiration (end expiration pause at 5 cm H₂O). This fraction includes the lung tissue that is “consolidated” (i.e., not openable at 45 cm H₂O airway pressure) and the tissue that is collapsed but openable at 45 cm H₂O. These values were chosen to produce a minimal risk during the maneuver and because this is the most frequently used in the literature for recruitment maneuver.⁵⁹ In patients with elevated chest wall elastance, the resulting transpulmonary pressure could be insufficient, however, to overcome the opening pressure of some pulmonary units.

The total fractions of nonaerated lung tissue (consolidated plus recruitable) and the recruitable tissue alone (tissue that regains aeration at 45 cm H₂O airway pressure) are strongly associated with mortality rate. The data of Figure 11.2 show the inadequacy of the term ALI/ARDS, as currently defined, to describe the lung injury.⁴⁸ The data refer to a population of ALI/ARDS patients, classified according to the American-European Consensus Conference on ARDS,⁴⁵ in which a CT-based quantitative analysis of the whole lung parenchyma at 5 cm H₂O PEEP was performed. As shown in Figure 11.3, the fraction of the nonaerated lung tissue (consolidated or collapsed or both) may range from 5% to 70% of the entire lung parenchyma. Patients meeting ALI/ARDS criteria may have a baby lung size very close to that of normal subjects, or a baby lung that is just a small fraction of the expected normal lung. When the distending force is applied to the lungs by the mechanical ventilator, previously collapsed lung regions may open. Lung recruitability may be expressed as the amount of lung tissue regaining aeration when increasing the applied driving force from 5 to 45 cm H₂O. As shown in Figure 11.3, in some patients, lung recruitability was almost negligible, whereas in others it was equal to 25% to 35% of the entire lung parenchyma. Lung recruitability was strongly associated with the fraction of nonaerated tissue, suggesting that the greater the inflammatory edema, the greater the lung collapse. It seems that the best way to assess the overall lung severity and the related lung recruitability, both strongly associated with mortality rate, is the CT scan analysis. Physiologic variables are poor indicators of the severity of the lung injury. This relationship is shown in Figure 11.4—for a large variation of lung injury, ranging from 20% to 60% of the lung parenchyma, the values of PaO₂/FIO₂ ratio, lung compliance, and PaCO₂ greatly overlap.