The Nature of Recruitment and De-Recruitment and Its Implications for Management of ARDS

Temporal Aspects of Lung Recruitment

Two temporal aspects influence the effectiveness of recruitment: (1) the duration of any particular recruitment maneuver itself, and (2) the clinician-set inspiratory time chosen during the maneuver. Some of what is discussed below reflects this ambiguity as to precisely what occurs when we observe recruitment. Some of this (but by no means all) has been clarified by the advent of lung parenchymal microimaging in animal models, as discussed above. The following 2 sections provide a historical narrative on the development of our understanding as well as the persistent ambiguity surrounding recruitment from the 1960s to the 1990s.

Creep: Fast Versus Slow Pulmonary Compartments

The term “creep” was coined in the 1960s to describe the progressive increase in volume over time when the lungs are subjected to “constant” pressure inflations.^60,61 More broadly referred to as hysteresis or stress adaptation, creep expresses how tissue, once deformed, resists returning to its former shape. This is attributed to adaptive surface tension forces in the lungs and intrinsic viscoelastic properties of both lung and chest wall tissue (eg, the presence of elastic fibers in smooth muscle, skeletal muscle, ligaments, and tendons) as well as abdominal organs.⁶¹

Under normal physiology, a 2-phase process consisting of fast and slow compartments was described in animals.⁶⁰ During a 10-s inflation hold, an initial rapid (2 s) phase is followed by a slow (8 s) phase of continued tissue stretching; the latter was attributed primarily to alveolar recruitment and reduced alveolar surface tension, and to a lesser degree alterations in tissue viscoelastic properties.⁶² Stress adaptation was directly associated with increasing driving pressures and reaching maximum creep at 33 cm H₂O. Increasing sustained inflation intensity (ie, 120 s at 33 and 39 cm H₂O) effected additional stress adaptation (Fig. 1).⁶⁰

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Association between increasing driving pressures and alveolar stress adaptation in an animal model with normal chest mechanics. Data from Reference 60.

Stress adaptation has been observed in anesthetized normal subjects undergoing step inflations (similar to constructing a pressure-volume curve), stabilizing at 5–7 s at different volumes.⁶¹ Two thirds of stress adaptation was attributed to the lungs, with the chest wall exhibiting a smaller, slower time course. This was ascribed to tissue properties in both structures rather than alveolar recruitment and gas redistribution.⁶¹

Evidence Supporting Creep Phenomenon in ARDS

In ARDS, slow volume changes following a 10 cm H₂O PEEP increase were first reported by Katz et al,⁴ whereby 67% of volume change occurred during the first breath and 90% by the fifth breath. The remaining increase occurred over 40 min and was attributed either to stress adaptation or alveolar recruitment. Similar to other findings,⁶¹ 62% of the changes were attributed to the lungs and 38% to the chest wall.⁴

Slowly distensible pulmonary compartments in severe ARDS have been reported by others.⁶³ Using a 5 cm H₂O PEEP increment, only 37% of subjects exhibited a slow recruitment compartment, whereas 79% displayed a slow de-recruitment compartment, consistent with other studies reporting delayed de-recruitment following step-decreases in PEEP.²⁶ The mean inflation time constant (τ) of the slow compartment was 9.4 ± 7.3 s. As 95% equilibration occurs at 3 τ and 99% at 5 τ,⁶⁴ recruitment (or stress adaptation) of slow pulmonary compartments would reach 95–99% volume equilibration at mean times of 28–47 s with an upper 95% confidence limit of 43–72 s. As a reference, in normal subjects under general anesthesia, atelectasis reversal occurs at τ of 2.6 s (95–99% reversal at 8–13 s).¹²

In contrast, when oxygenation is the variable of interest, the temporal impact on recruitment is exaggerated. Several studies examined the time required to establish steady state oxygenation in ARDS following a PEEP increase or after initiating sigh breaths.^27,65,66 Setting PEEP above the lower inflection point (PEEP 14 ± 3 cm H₂O), 90% of maximal improvement occurred at 20 ± 19 min.⁶⁵ However, in another study, a 10 cm H₂O PEEP increase produced no apparent oxygenation plateau (ie, progressively increased by 5–60 min).⁶⁶ When augmenting LPV with intermittent sigh breaths (P_plat 45 cm H₂O; PEEP 14 cm H₂O) maximum improvements occurred at 30 min for both and end-expiratory lung volume.²⁷

An intriguing aspect of recruitment are transient (pulmonary) states observed in ARDS when the ventilatory pattern was altered.⁶⁷ Prolonged effects of recruitment were noted after various manipulations, including a single PEEP step, a PEEP wave maneuver, and an undulating PEEP pattern. One hour following a PEEP increase from 13 to 21 cm H₂O, FRC rose 150% greater than that predicted by C_RS of the fast pulmonary compartment (ie, baby lung).⁶⁸ These results were similar to those reported by Katz et al.⁴

In the PEEP wave study, a brief repetitive cycle of incremental ascending and descending PEEP with a maximum PEEP change of 10 cm H₂O was repeated 5 times over several hours. When PEEP was returned to the initial settings, stabilized at 10 mm Hg above the previous baseline. The phenomenon occurred with each successive PEEP wave so that, at the end of the experimental run, was 80 mm Hg higher than at the initial baseline.⁶⁷

The undulating PEEP study assessed the upper limit of time constant distributions using a PEEP cycle above and below a baseline PEEP of 14 cm H₂O. PEEP was titrated in increments of 7 cm H₂O up to 29 cm H₂O and down to 0 cm H₂O over 9 h. FRC measured 1 h following any PEEP change did not indicate a steady state in terms of recruitment or de-recruitment. The overall impression was that “the length of individual time constants in ARDS may exist in the region of hours.”⁶⁷

Slow, progressive lung recruitment frequently observed during prone position therapy supports the existence of slow pulmonary compartments in ARDS.⁶⁹ Initial improvement in oxygenation typically occurred within 30–60 min, yet it is not uncommon for improvements to become apparent only after 6 h, with continuing improvements sometimes observed over 20–36 h.⁶⁹ Prolonged recruitment maneuver (ie, 6–14 h) reversing profound refractory hypoxemia has been reported anecdotally in ARDS complicated by abdominal compartment syndrome⁷⁰ and in pronounced obesity when combined with prone position.⁶⁹

These findings underscore the considerable difference in the time frames chosen to evaluate oxygenation response following a recruitment maneuver. Several recruitment maneuver studies that will be discussed in the next section used 2-min equilibration periods between all or some of the PEEP steps,^17,40,71-73 which is consistent with classic physiologic studies.^60,67,74 While the 2-min limit allows for stress adaptation, it also limits exposure to severe respiratory acidemia⁴⁰ and potential cardiovascular instability from alterations in right and left ventricular function.^75-78 Most importantly, these necessary time constraints imposed by very high pressure recruitment maneuver techniques limits our ability to fully understand the actual recruitment potential in ARDS.

Selecting Inspiratory Time During Recruitment Maneuver

The other temporal aspect is whether the inspiratory time per breath impacts the overall effectiveness of a recruitment maneuver. This is likely dependent upon whether the clinician-set inspiratory time is appropriate for the inspiratory time constant of individual patients. In general, ARDS subjects have an inspiratory τ of 0.17–0.41 s,² which (depending upon syndrome severity) would result in 95% and 99% estimated equilibration between airway and alveolar pressures at ∼ 0.5–1.2 s and 0.9–2.1 s, respectively. However, these estimates are based on assumptions of mono-exponential functions of constant elastances and resistances throughout inspiration, and therefore neglect the impact of mechanical inhomogeneity in ARDS.⁷⁹ Moreover, they ignore the impact of continued gas mixing and redistribution (ie, pendelluft motion) and increased diffusion time on both oxygenation and dead-space ventilation by which recruitment maneuver efficacy often is assessed.

The range of inspiratory time reported in recruitment maneuver studies have varied: 1.5 s (single PEEP step),⁶³ 2.5 ± 1.1 s (for sigh breaths), ²⁷ 2–3 s^17,71,72,80 (for staircase recruitment maneuver studies or unspecified),⁴⁰ whereas others used an end-inspiratory pause of 5–7 s.^34,81 In a lung lavage model of acute lung injury, in vivo microscopic studies of subpleural alveoli during a recruitment maneuver at 40 cm H₂O found that, over a period of 40 s, ∼ 85% of recruitment occurred by 2 s.⁸² These data support clinical use of inspiratory times of 2–3 s to maximize per-breath recruitment potential during an recruitment maneuver.

Distribution of TOP and Recruitment

Small physiologic studies suggest varying degrees of recruitment occur throughout the lung. An early CT study reported that potentially recruitable lung in moderate-to-severe ARDS averaged 21 ± 10% and required a P_plat of 45 cm H₂O (whereas ∼ 25% remained collapsed).⁸⁰ Also, there may exist nodal points whereby full recruitment transitions down the lungs from mid-to-dorsal regions when P_plat of 30, 35, and 45 cm H₂O are reached (with the least amount of nonaerated tissue observed at 45 cm H₂O).^40,85 Subsequent studies have reaffirmed that TOP varies down the ventral-dorsal axis in ARDS. Upper zones had a negligible TOP of 0–4 cm H₂O, whereas middle zones had a TOP of 4–7 cm H₂O and dorsal lung recruitment commenced at ∼ 20 cm H₂O.⁸¹

Similar to initial post-natal breaths, achieving a TOP in ARDS is not synonymous with full recruitment. Early pressure-volume curve studies of ARDS interpreted the lower inflection point as the TOP needed to recruit collapsed peripheral airways and alveoli, but it was misconstrued as the anchoring point for setting best PEEP.⁸⁹ It later became apparent that recruitment merely commenced in the upper lung zones at the lower inflection point and continued throughout the inspiratory pressure-volume limb.^90,91 Likewise, despite TOPs of 4–7 cm H₂O (middle) and ∼ 20 cm H₂O (dorsal), maximum recruitment in these regions occurs at 20–30 cm H₂O and 45 cm H₂O, respectively.^81,85

Other CT imaging studies reported that nonaerated lung tissue progressively decreased from 55% (at a baseline ventilation with 10 cm H₂O of PEEP) to 23% at a P_plat of 40 cm H₂O and to 10% at a P_plat of 50 cm H₂O.⁷¹ Improved recruitment was observed even when P_plat increased from a baseline of 28–32 cm H₂O to 36–41 cm H₂O at essentially the same PEEP level.⁵ Most importantly (in light of the ART study results), extending a recruitment maneuver to a P_plat of 60 cm H₂O only reduced nonaerated tissue by 5%. Full recruitment has been reported at P_plat of 40–51 cm H₂O,⁷² whereas others have reported increasing percentages of subjects achieving full recruitment as P_plat increased: 46% at 40 cm H₂O, ∼ 60% at 45 cm H₂O, and ∼ 70% at 50 cm H₂O).⁴⁰ In the seminal study on the time course of FRC improvement with PEEP in subjects with acute respiratory failure, the majority of whom likely would have met the current definition of ARDS, a P_plat of 40 cm H₂O and PEEP of 18 cm H₂O were needed to return FRC to normal.⁴ Thus, in the context of refractory hypoxemia even at moderate to high levels of PEEP, a recruitment maneuver P_plat of at least 40 cm H₂O probably should be targeted.

Furthermore, there is speculation that the recruitable lung represents a penumbra of inflamed tissue surrounding a core nidus of compartmentalized injury, constituting a mixture of collapsed or partially flooded air spaces.^21,92 The remaining ∼ 25% of nonaerated lung tissue, despite recruiting pressures of 45 cm H₂O,⁸⁰ likely signifies consolidated tissue, at least in those with normal body habitus (see below).

Interpretive Limitations of Mechanistic Studies of Recruitment

Interpreting these recruitment maneuver studies raises several issues. First, confounding factors influence the potential effectiveness of recruitment maneuvers in ARDS. These may include: (1) the inherently heterogenous nature and unique spatial patterns of acute lung injury among individual patients; (2) apparent differences in the response to recruitment maneuver related either to initiating pathways (direct vs indirect, interstitial vs alveolar edema) or the severity of injury (eg, the degree of inflammation and magnitude of edema formation); (3) timing of recruitment maneuver relative to syndrome onset; (4) the ventilatory strategy used prior to initiating recruitment maneuver; (5) alterations in chest wall mechanics; and (6) hemodynamic status (eg, various vasoactive drugs that might affect cardiac output and pulmonary blood flow distribution).^24,93-98 Moreover, mechanistic recruitment maneuver studies require highly complex, clinically impractical methodologies that limit the number of subjects who can be studied and thus limits the generalizability of results to individual patients.

Second, variables chosen to signify full recruitment differed between studies, which introduces interpretative ambiguity. Borges et al⁴⁰ used + > 400 mm Hg. Adjusting for the range of mean across recruitment maneuver steps (70–95 mm Hg) yields a corresponding of ∼ 300–330 mm Hg. Povoa et al⁷² used a of 250 mm Hg on an of 1, and de Matos et al⁷¹ reported the P_plat at which nonaerated alveoli was minimal. Studies by both Crotti et al⁸⁵ and Caironi et al⁹² merely reported the degree of recruitment observed at a fixed P_plat of 45 cm H₂O as a surrogate measure of total lung capacity.

Third, over the years, recruitment maneuver studies have utilized different measurement techniques that have influenced both the results and their interpretation.³² Recruitment inferred from chest mechanics (eg, change in end-expiratory lung volume measured during construction of pressure-volume curves or after step changes in PEEP) reflects increased aeration of partially and fully inflated alveoli, as well as recruitment of previously collapsed or noncommunicating alveoli. In this review, we have focused on CT-based studies. Although ambiguities exist with this technique, it nonetheless provides a high degree of differentiation between non-, poorly and well-inflated alveoli (see below). As would be anticipated, recruitment inferred from chest mechanics analysis estimate much greater recruitment than those based on CT analysis.³²

Taken together, these potential confounding variables (ie. relatively small numbers of study subjects, variations in both technique and primary endpoints) limit the generalizability of recruitment maneuver study results to individual patients, let alone navigating the contentious discourse regarding their interpretation.

Sponge Theory and Superimposed Hydrostatic Pressures

Setting aside common clinically induced causes (eg, circuit disconnection, endotracheal suctioning), lung de-recruitment in ARDS is thought to be caused largely by super-imposed hydrostatic pressure of overlying edematous lung tissue and mediastinal structures, as well as increased weight of the chest wall (eg, thoracic anasarca, ascites).⁹⁴ This is based upon the sponge theory posited to explain rapid redistribution of lung densities on CT scans from dorsal to ventral regions during placement in prone position.⁷⁶ Two facts support the notion that this represents a shift in gravitational forces applied to the lungs. First, overall lung density was unchanged, suggesting lung tissue mass (ie, edema, blood, cellular content or debris) had remained stable.⁹⁹ Second, although pulmonary edema clearance in ARDS is severely impaired (6%/h),¹⁰⁰ edema fluid is removed through the lymphatic system and does not freely redistribute through lung tissue.⁹⁹

PEEP and De-Recruitment

A recurring and relatively uniform finding in many of the early recruitment maneuver studies was that, when ventilation was resumed at the previous PEEP level, oxygenation improvements dissipated rapidly over time despite relatively high baseline PEEP (∼ 12–15 cm H₂O).^{22,23,27,101,102} In contrast, oxygenation improvements could be sustained following recruitment maneuver when higher post-recruitment maneuver PEEP levels were maintained (eg, ∼ 6–7 cm H₂O above baseline).²¹ Acute lung injury models also reported that oxygenation after a recruitment maneuver was PEEP-dependent, with the highest sustained improvement occurring at PEEP of 16 cm H₂O (vs 12 or 8 cm H₂O).⁷⁷ That the sustained improvement was independent of recruitment maneuver methodology suggests recruitment and de-recruitment occur through different mechanisms.

When pleural pressure exceeds alveolar pressure at end-expiration, de-recruitment occurs over time irrespective of previous volume history.⁷³ In ARDS, de-recruitment is a continuous process that becomes prominent at PEEP < 15 cm H₂O.⁸⁵ De-recruitment appears to cease in the upper and hilar lung zones at PEEP of 10 cm H₂O, whereas it continues in dorsal regions, reaching a maximum collapse rate at 5 cm H₂O.⁸⁵ De-recruitment modeling suggests the speed of collapse also increases as PEEP decreases.⁵⁹ Similarly, a decremental PEEP study noted that pleural pressure exceeded alveolar pressure once PEEP decreased below ∼ 9 ± 5 cm H₂O, whereas in some subjects de-recruitment occurred at PEEP < 20 cm H₂O.⁷³

These findings suggest 3 potential PEEP targets that might reduce de-recruitment during the acute phase of ARDS: (1) minimum PEEP of 10–12 cm H₂O, (2) a general target of 16 cm H₂O, and (3) ≥ 20 cm H₂O in very severe cases, particularly those with reduced chest wall compliance. This is similar to the better PEEP strategy proposed by Gattinoni and colleagues.¹⁰³

Superimposed Pressure, De-Recruitment, and ARDS Severity

The largest and perhaps most comprehensive CT study reported the maximum range of ventral-dorsal superimposed hydrostatic pressure was 6–18 cm H₂O.⁹⁴ Interestingly, the mean hydrostatic pressure was similar between Berlin classifications of mild, moderate, and severe ARDS (12 ± 3, 12 ± 2, 13 ± 1 cm H₂O, respectively, P = .053). Factoring in PEEP required to counter both superimposed hydrostatic pressure and chest wall elastance yielded PEEP estimates of 16 ± 8, 16 ± 5, and 18 ± 5 cm H₂O, respectively (P = .48).

A particularly interesting finding was that PEEP requirements did not differ between those characterized as having low or high recruitment potential and based on the observation that maximum superimposed hydrostatic pressure between the 2 groups differed by only 1–2 cm H₂O. Thus, neither superimposed hydrostatic pressure nor chest wall elastance correlated appreciably with recruitment potential.

This implies that superimposed hydrostatic pressure enters the calculus of setting PEEP to preserve lung stability following recruitment rather than causing recruitment. These observations led the authors to dissuade clinicians from reflexively treating low recruitability (ie, lobar ARDS) with PEEP levels of ∼ 15 cm H₂O simply to prevent shear injury in “a few grams of lung tissue,” given the greater risk of hemodynamic compromise and regional overdistention in middle and ventral lung zones.⁹⁴

Elevated Intra-Abdominal Pressure in ARDS

The ventral-dorsal pleural pressure gradient in the supine position determines resting alveolar size and largely reflects gravitational forces imposed by the abdomen, which is a more dense, fluid-like compartment with a volume twice that of an air-filled thorax.^104,105 Normal intra-abdominal pressure (IAP) is ∼ 5–7 mm Hg (7–10 cm H₂O), whereas intra-abdominal hypertension is defined as IAP > 12 mm Hg (16 cm H₂O) with 20–60% pressure transmission to the thorax.¹⁰⁶ Therefore, severe hypoxemia coinciding with intra-abdominal hypertension is a compelling indication for OLV.

Thoracoabdominal Mechanics, De-Recruitment, and Intra-Abdominal Hypertension

Elevated IAP displaces the diaphragm cephalad into the thorax and stiffens the abdominal portion of the chest wall, such that pleural pressure becomes more positive. This is particularly acute in the dorsal-caudal regions in the supine position, causing reduced lung and chest wall compliance, increased tissue and airways resistance, and compressive atelectasis.^107-109 Under these conditions, alveolar de-recruitment from tissue compression (vs alveolar consolidation) is more likely the primary cause of refractory hypoxemia, hence a recruitment maneuver is more likely to be effective.

Abdominal compartment syndrome (IAP > 25 mm Hg; > 34 cm H₂O)¹¹⁰ is associated with substantial nonaerated and poorly aerated lung tissue (23% and 18%, respectively).¹¹¹ At these extraordinary pressures, respiratory system inertance, normally considered negligible, may become significant and therefore would increase TOP.¹¹² Inertance refers to the acceleration of gas molecules as well as displacement of resting lung and chest wall tissues, including the abdominal contents.¹⁰⁹

During quiet breathing with normal body habitus, inertance accounts for < 5% of driving pressure.¹¹³ In morbid obesity, inertance is ∼ 4-fold higher, and up to 68% can be accounted for by chest wall tissue.¹⁰⁹ Although its relevance to ARDS is unknown, it is notable that, in morbidly obese subjects, the driving pressure required to overcome inertance alone during maximal ventilation maneuvers reaches 40 cm H₂O.¹⁰⁹ In a case of ARDS and abdominal compartment syndrome, a similar driving pressure (P_plat 80 cm H₂O and PEEP 45–50 cm H₂O) was required to increase from 23 to 350 mm Hg when surgical decompression could not be attempted.⁷⁰

Intra-Abdominal Hypertension and ARDS

Intra-abdominal hypertension is common in severe ARDS¹⁰⁸ and is particularly prevalent in extrapulmonary cases.⁴³ It occurs in pulmonary ARDS complicated by morbid obesity (ie, mass loading), where IAP is ∼ 12–19 cm H₂O,¹¹⁴ as well as other conditions such as ascites from abdominal sepsis, pancreatitis, or hepatic failure.^115,116 In acute lung injury models, IAP of 20 cm H₂O greatly exacerbated pulmonary edema formation and increased intrapulmonary shunting.^117,118 Mean IAP of ∼ 22 cm H₂O^43,116,119 and end-expiratory esophageal pressures of ∼ 20 cm H₂O have been reported in cases of severe ARDS.¹²⁰

IAP is particularly relevant in treating refractory hypoxemia. A preclinical study reported that, at IAP of 24–35 cm H₂O, high PEEP (ie, 15 cm H₂O) was equally ineffective as low to moderate PEEP (ie, 5–12 cm H₂O) in improving FRC and .¹²¹ This led to a follow-up study of IAP-matching PEEP in acute lung injury with intra-abdominal hypertension (ie, 16–25 cm H₂O). Higher levels of P_plat and PEEP used in the pressure control ventilation recruitment maneuver strategies described above were needed to improve FRC and oxygenation (Fig. 4).¹¹⁷

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Relationship between PEEP, plateau pressure (P_plat) and chest wall elastance (E_cw) at increasing levels of intra-abdominal pressure in an animal model of acute lung injury. Data from Reference 118.

Adding half of the measured IAP to the recruitment maneuver pressure targets has been suggested.¹⁰⁶ For example, a recruitment maneuver of 45 cm H₂O⁹² applied to IAP representing conditions of intra-abdominal hypertension (16 cm H₂O), average IAP in reported in ARDS (22 cm H₂O) or severe abdominal compartment syndrome (≥ 50 cm H₂O)^110,122 would require adjusting P_plat upward to 53, 56, and 70 cm H₂O, respectively.

Attempting a recruitment maneuver in a patient with intra-abdominal hypertension requires assessing overall risk/benefit ratio. Elevated pleural and intra-abdominal pressures impede hemodynamic function and lymphatic drainage and therefore carries the risk of worsening both pulmonary edema and intra-abdominal hypertension as well as risking hemodynamic collapse.¹²³ In the context of abdominal compartment syndrome, it should probably be considered only when surgical decompression carries an even greater risk.

Impact of PEEP on Volume Distribution in ARDS

Finally, regardless of P_plat or PEEP, gas distribution in ARDS steadily decreases down the ventral-dorsal axis with an upper-to-lower lung volume distribution ratio of 2.2:1 at ambient end-expiratory pressure. At PEEP of 20 cm H₂O ventral-dorsal gas distribution was essentially equivalent (1.1:1).⁸¹ For the dorsal regions (ie, those having the greatest impact on gas exchange) this volume redistribution translated into increased end-expiratory lung volumes from ∼ 10% to 25% and increased end-inspiratory lung volumes from 15% to 35%. These findings were supported by an electrical impedance tomography study of OLV wherein ventral/dorsal tidal volume ratio decreased from 2.01 ± 0.36 to 1.19 ± 0.10 (P < .01).¹²⁴

Radiologic Factors

CT scans are the accepted standard for evaluating topographic distribution of aerated and nonaerated lung tissue in ARDS, inferred by the lungs ability to attenuate x-rays.¹²⁵ The radiologic definition of consolidation is markedly increased lung attenuation obscuring pulmonary vessels caused by atelectasis or alveolar filling, whereas in pathology the term specifically refers to the latter.¹²⁶ Attenuation is measured by the Hounsfield linear density scale that assigns a numeric value (Hounsfield units [HU]) differentiating between bone (+1,000 HU), water (0 HU) and air (–1,000 HU).¹²⁷ Values between these 3 points are used to convey various states of pulmonary tissue, with values between −100 HU and +100 HU considered to represent collapsed tissue (Table 2).^12,81,94,128

From these interpretations, pulmonary gas-tissue ratios are calculated and used to infer the response to recruitment maneuver and PEEP. Yet, the designation of lung “tissue” also includes extravascular fluid and blood.¹²⁷ Thus, CT imaging represents “the quantity of air being introduced into a diseased lung,” hence the statement, “one pixel is not an alveolus.”¹¹⁵ Lung CT imaging interpretation relies upon an unprovable assumption of homogenous alveolar filling in condensed lung tissue, whereas in reality it likely includes already aerated alveoli.^56,115 In addition, estimating the reduction in nonaerated tissue is dependent upon the number of CT sections sampled (compared to whole lung scans). For example, a single juxta-diaphragmatic section may result in either over- or underestimation of recruitment, whereas adding samples of apical and hilar regions tends to overestimate recruitment.¹²⁹

In spite of the strong association found between radiologic assessment of alveolar recruitment and oxygenation,^40,130 a complex interaction of other factors contributes to improved oxygenation (eg, increased ventilation-perfusion matching,¹³¹ decreased cardiac output with redistribution of pulmonary perfusion,¹³² reduced edema formation,¹³³ and its redistribution to the perivascular spaces^56,134). Skeptics claim radiologic evidence supporting lung recruitment are “inferences about alveolar micromechanics from measurements made on a scale several orders of magnitude greater than that of the structures of interest.”¹³⁵ The volume element of a CT image (ie, the voxel) is ∼ 2−2.6 mm³,^32,127 whereas a single alveolus is ∼ 0.12 mm³.¹³⁶ Thus, a single voxel may represent a tissue section consisting of ∼ 15 discrete alveoli.

The importance of this limitation becomes apparent in lung microimaging of gas dynamics within alveolar clusters. Animal models of acute lung injury observed pronounced pendelluft motion between adjacent alveoli (some slowly inflating during expiration, some deflating during inspiration), as well as paradoxically simultaneous recruitment and de-recruitment, while still others either synchronously inflate and deflate or appear stunned (ie, remaining motionless at a constant volume).^137,138 This localized inter-alveolar asynchrony and instability results from mechanical interdependency between neighboring alveoli and increases with the severity of injury.¹³⁸ Although CT imaging studies provide invaluable information on the nature of recruitment and de-recruitment, they are clinically impractical for routine use; in addition, there remains assumptive ambiguity and therefore a risk of over-interpretation.

Rheologic Factors

During expiration, distal airway de-recruitment occurs as increasing surface tension causes liquid bridges to reform, drawing airway and alveolar walls together.^56,139 An in vivo study of acute lung injury confirmed the presence of liquid menisci forming dense bridges across small peripheral airways.¹³⁸ Therefore, the perception of alveolar recruitment in acutely injured lungs may be explained as the breaking of foam bridges and displacement of pulmonary edema fluid, resulting in increased alveolar ventilation.^56,135,140 Thus, other factors determine the force required to re-open the lungs: surface-tension forces (accounting for 50–60% of lung elastance), the presence of biologically active surfactant (in both alveoli and distal airways), viscosity and thickness of airway edema, and overcoming strain energy in collapsed small airways (see below).^{48-50,56,139,141}

Even sponge model proponents acknowledge that compression atelectasis likely represents a mixture of alveolar and small airway collapse.⁸¹ What remains undisputed is that specific and reproducible ranges of airway pressures transmitted to the lung parenchyma are required to improve regional aeration and gas exchange in ARDS, and that the recruitment of collapsed or obstructed airways and alveoli invariably involves epithelial cell deformation and therefore likely causes shear injury^139,140 and exacerbates baseline airway epithelial injury associated with ARDS.⁴⁵ Greater injury is thought to occur with reopening collapsed versus obstructed airways.¹⁴⁰

Histopathologic Factors

Ambiguity surrounding the effectiveness and appropriateness of recruitment maneuvers partly depends upon whether atelectasis (ie, degassed alveoli), intra-alveolar edema (ie, flooded alveolar units and peripheral airways), or interstitial edema is the predominant lesion causing refractory hypoxemia, as well as the intensity of edema.⁹⁸ Historically the most prominent autopsy findings in early ARDS included some combination of interstitial and alveolar edema or hemorrhage and hyaline membranes,^142-152 along with a substantial subset reporting atelectasis.^{144,145,147,149,153} In what eventually would be called ARDS, the term congestive atelectasis was used to describe “diffuse non-obstructive collapse of pulmonary alveoli and intense interstitial edema and pulmonary capillary congestion,”¹⁴⁷ leading to excessive surface tension forces causing collapse.¹⁴⁸ More recently, this has been redefined as inflammatory (ie, congestive) atelectasis versus compression atelectasis.¹⁵⁴

These characteristics defined diffuse alveolar damage, the histopathologic hallmark of ARDS.¹⁴² During the first week of ARDS confirmed with diffuse alveolar damage, intra-alveolar edema tends to be highest (90% of cases) but remains prevalent during subsequent weeks (74% of cases).¹⁵⁵ Only ∼ 50% of ARDS cases now present with diffuse alveolar damage,¹⁵⁶ its decrease coinciding with the emergence of LPV.¹⁵⁷ ARDS without diffuse alveolar damage has been associated primarily associated with pneumonia. This is characterized less by intense interstitial edema and alveolar neutrophil infiltration localized in the terminal bronchioles.^156,158

A study that matched PEEP responsiveness to lung biopsy and autopsy samples reported that subjects exhibiting minimal oxygenation response had complete alveolar filling with purulent or hemorrhagic material. Those exhibiting the greatest oxygenation response had less intense alveolar edema and were distinguished by hyaline membrane formation, interstitial edema, and atelectasis.¹⁵⁹

Direct Versus Indirect Injury and Injury Severity

Both direct and indirect forms of ARDS include alveolar collapse,⁴² yet direct (ie, alveolar epithelial) injury has been characterized more by intense collapse and alveolar edema but minimal interstitial edema, whereas indirect (ie, capillary endothelial) injury is associated with more intense interstitial edema than alveolar edema.¹⁶⁰ Similar findings were reported in other studies.¹²⁶ Moreover, direct injury has been associated with higher pulmonary microvascular permeability that, over a period of days, coincides with higher levels of extravascular lung water.¹⁶¹ Some evidence suggests that recruitment maneuver (at least when using the sustained inflation technique) might be ineffective in the presence of high extravascular lung water (∼ 16 mL/kg).⁹⁸

Inconsistencies between histologic findings in ARDS likely have many sources due to the limited number of samples, the heterogeneous nature of ARDS and associated lesions, and its timing relative to syndrome onset. Irrespective of these, such inconsistencies suggest that simplistic conceptual models guiding recruitment maneuver have limited utility because the varied histopathologic changes in ARDS coexist across a spectrum and that lesions evolve over time.¹⁵⁹ Moreover, direct injury from pneumonia disrupting alveolar membrane integrity (ie, loss of bacterial compartmentalization) can induce indirect, secondary injury to noninfected lung regions through systemic cytokine release.¹⁶² In fact, a substantial number of subjects with ARDS with either aspiration or pneumonia as primary etiology also have sepsis as a secondary source of lung injury (20% and 40%, respectively).¹⁶³

Furthermore, secondary analysis of several recruitment maneuver CT studies concluded that recruitment is likely determined more by the severity of injury and corresponding edema formation than injury mechanism per se.⁹³ As lung injury severity increases, so too does the degree of pulmonary capillary permeability and the magnitude of extravascular lung water.¹⁶⁴ In general, regardless of injury mechanisms, greater recruitment potential is present in ARDS characterized by diffuse versus predominant dorsal opacities.^101,165-167 Unfortunately, this is not a distinction that can be made by clinicians when chest radiographs are the only practical tool available when contemplating whether to pursue treating refractory hypoxemia with recruitment maneuver.

In a secondary analysis of recruitment maneuver studies evaluated with CT, estimates of recruitability were actually higher in direct injury.⁹³ Several factors were cited that provide important insights into the interpretation of recruitment maneuver studies. First, the timing of recruitment maneuver relative to ARDS onset influences recruitability. Over time, edema fluid is slowly reabsorbed while concurrently fibrotic and tissue repair mechanisms evolve. Second, the duration and fidelity to LPV prior to initiating a recruitment maneuver will influence recruitment potential regardless of injury mechanism. Third, the higher correlation between direct injury and ARDS severity may reflect the degree of bacterial diffusion throughout the lung parenchyma.⁹³ This in turn suggests relatively greater consolidation in direct injury (with a corresponding brisk reactive edema formation) possibly producing greater edema than that caused by distant organ injury. However, the investigators stressed that extrapulmonary injury or infection can cause equivalent severity. Thus, when only a small number of subjects are studied (ie, selection bias), the results may suggest equivalence, relatively greater, or relatively lesser lung recruitability between direct versus indirect injury.

Optimizing Oxygenation and Minimizing Risk of Atelectrauma

FRC represents the alveolar volume and is the primary determinant of .¹⁷⁷ Therefore, increased in response to increased PEEP or recruitment maneuver is a bedside convenience to infer changes in FRC and, by extension, shear injury risk. Unfortunately, the logic linking these 3 phenomena is precarious.

Depending upon P_plat, a substantial portion of early FRC increase (ie, the fast pulmonary compartment) represents expansion of normally inflated or underinflated alveoli and not recruitment.⁴ In addition, arbitrary thresholds used to signify full recruitment (ie, 250–330 mm Hg)^40,72 are literally false. Full recruitment implies normal pulmonary oxygen diffusion function (eg, ≥ 450 mm Hg). This does not occur in ARDS because of varying degrees of tissue consolidation and slow resolution of pulmonary edema. Thus, the thresholds of 250–330 mm Hg used to evaluate recruitment maneuver effectiveness suggest a tacit acknowledgment that the term full recruitment is meant figuratively.

Beyond these vagaries lies the crux of the debate: Does OLV materially reduce the risk of repetitive shear injury compared to traditional LPV? This is unlikely for the majority of ARDS cases. First, a P_plat of ≥ 40 cm H₂O is needed to reopen distal airways and alveoli deep within the dorsal lung; these units remain closed and protected from shear injury when P_plat is limited to ≤ 30 cm H₂O. Second, in ARDS substantial portions of lung tissue appear to reach full recruitment at P_plat ≤ 30 cm H₂O, and its stability appears to be maintained when PEEP is set at 10–15 cm H₂O. In addition, evidence from several preclinical studies suggests that atelectatic areas are relatively protected from shear injury by intra-alveolar edema, with most damage caused by excessive stress developed in the peripheral airways.¹⁷⁸ In these studies, tidal overdistention was a more important contributor to pro-inflammatory cytokine expression than shear injury. Third, microimaging of subpleural alveoli in acute lung injury models revealed that, despite stable levels of driving pressure and PEEP, there exist patterns of recruitment and de-recruitment between interdependent alveoli, even at high PEEP levels, that appear to fluctuate minute by minute.¹³⁸ Thus, the notion of eliminating de-recruitment and atelectrauma in ARDS appears illusory.

Implications of Slow Pulmonary Compartments

Integrating the temporal issues involved in recruitment, with evidence that most recruitment occurs at ≤ 50 cm H₂O, and the increased mortality risk reported in the ART study,¹⁷ it behooves us to reflect upon the need for a recruitment maneuver and how it might be approached going forward. Compelling evidence of slow pulmonary compartments in ARDS is at odds with the current recruitment maneuver strategy and raises questions of whether brief recruitment periods reflect the actual effectiveness of a specific P_plat. By extension, this influences the decision to use higher pressures with increasing risk of injury and hemodynamic compromise. Moreover, limited intensity recruitment maneuver studies such as the extended sigh, the prolonged, and the slow moderate recruitment maneuvers cited above all observed substantial recruitment at pressures ≤ 40 cm H₂O over a period of several minutes.^33-35,37,38 To date, no study has investigated whether an extended trial of super-PEEP limited to 25–30 cm H₂O and driving pressures of 15 cm H₂O might provide sufficiently stable oxygenation over a period of several hours.

Also, the relative importance of using an inspiratory time of 2–3 s during a recruitment maneuver, while supported by preclinical data, has not been evaluated clinically. This strategy substantially limits recruitment maneuvers because it restricts minute ventilation in more severe manifestations of ARDS that are associated with highly elevated physiologic dead space¹⁷⁹ and places additional strain on right-ventricular function.^78,170 In preclinical studies, an inspiratory time of 1.4 s is generally sufficient for recruiting the fast pulmonary compartment.¹⁸⁰ In light of studies describing transient [pulmonary] states, as well as those on prone positioning, it is worth considering whether more clinically appropriate inspiratory times used during LPV, if not optimal, might be sufficient to effect sufficient recruitment over time to reach oxygenation goals.