Perioperative Pulmonary Atelectasis: Part I. Biology and Mechanisms

Bronchioles and alveoli walls are composed of cells and extracellular matrix and covered by a liquid film on their luminal side containing surfactant. Each of these elements are exposed to expanding and collapsing forces.

Normal stress is the force per unit of area (A) perpendicular to the surface where the force is exerted. Three main components of the normal stresses acting on bronchioles and alveolar walls determine their expansion (fig. 1)⁴ : fluid pressure, tethering stress, and surface tension. A conceptual note on the physical meaning of these mechanical components is that, although related, they are not equivalent as pressure is a scalar (a physical quantity having only magnitude), whereas stress is a vector (a physical quantity with direction and magnitude).

• (1) Fluid pressure represents the pressure applied by fluids (gas or liquid) to the surface of the alveolar or bronchiolar wall. The net result of the fluid pressures derives from the difference of inside (P_i) and outside (P_o) pressures, expressed by the formulation of transmural pressure:

• (2) Tethering stress represents radial stresses caused by attachments of bronchioles and alveolar walls to adjacent structures through the tissue matrix. The radial tethering stress is mechanically transmitted to alveoli, bronchioles, and pleural surface through a network of collagen and elastin fibers composing the extracellular matrix in the pulmonary septa. These fibers are the force-bearing elements. The parenchymal cells themselves (epithelial and endothelial cells) have a lower mechanical contribution. Preserved lung interstitial architecture thus ensures the transmission of the tethering stress inside the lung parenchyma.⁵  The net effect of those stresses applied inside (F_i/A) and outside (F_o/A) the bronchioles or alveolar walls can be expressed as (fig. 1):

• (3) Surface tension represents the inward-acting radial stress arising from the circumferential components of forces applied by the thin layer of fluid lining the bronchioloalveolar walls resulting from the effect of the surface tension (T). For a spherical structure with radius R (alveoli), the Young–Laplace pressure equation expresses the relationship between the pressure difference across the fluid interface (P_w) and the surface tension as:

For a cylindrical structure of radius R (bronchioles), the relationship is:

Accordingly, the pressure difference across the fluid interface becomes substantial for small R. The surface tension is likely more important in the cylindric bronchioles than in alveoli, which are not strictly spherical.⁶  Pulmonary surfactant, a lipoprotein complex secreted by type II alveolar epithelial cells, is a critical biomechanical stabilizer to bronchioles and alveoli expansion.⁷  Its presence at the air–liquid interface strongly reduces surface tension, decreasing the magnitude of this collapsing contribution.

The balance of these forces and pressures allow for quantitative relationships in specific conditions (fig. 1):

• •

  In equilibrium, the balance of the expanding and collapsing radial stresses acting on bronchioles or alveolar walls should be zero⁴ :

• •

  No external tissue attachments are present at the pleural surface. Consequently, F_o = 0 and transmural pressure is determined by P_i (= alveolar pressure, P_alv) and P_o (= pleural pressure, P_pl). Radial stresses corresponding to the sum of inward-acting tissue and surface forces are balanced by the transmural pressure (P_alv – P_pl) acting on the pleural surface area (A_pl). Because the radius of the pleural surface curvature is large, the effect of surface tension is negligible (P_w ≈ 0). Accordingly, the balance of forces at the pleural surface is:

• •

  Within the lung, if all airways are patent, pressures on the two sides of alveolar walls (i.e., P_i and P_o) equal the same alveolar pressure (P_alv). Therefore, transmural pressure between adjacent alveoli is null, and the outward-acting tethering force (F_o) counteracts inward-acting tissue and surface forces:

Lung elastic recoil represents the propensity of lung tissue to shrink and is the main physiologic mechanism of passive exhalation. It results from the combined effects of: (1) extracellular matrix elastic fibers (contributing to F_i) and (2) bronchioloalveolar surface tension.^4,7  Accordingly, degradation of the elastic fibers in the extracellular matrix, as during emphysema, reduces elastic recoil, thus reducing the expiratory capacity and acting against alveolar collapse.⁸  Conversely, diseases leading to quantitative or qualitative surfactant impairment increase surface tension and facilitate alveolar collapse.⁷ 

Pulmonary interdependence represents the interplay of mechanical forces among lung tissue components: alveolar units, airways, vasculature, and extracellular matrix. For instance, interdependence during lung expansion transmits tethering stress to traction airway walls outwards.^5,9–11  Interdependence relies on normal lung architecture, including the extracellular matrix fibers.¹²  In a homogeneous lung, outward tethering stresses (ΣF_o/A_pl) are transmitted from the visceral pleura surface to the innermost lung regions. These stresses are determined by the elastic recoil pressure of the lung P_el(L),¹³  equal to the transmural pressure at the pleural surface:

A positive P_el(L), transmitted to the inner lung through interdependence, is the primary determinant of lung expansion. During awake spontaneous breathing, P_pl is negative throughout the pleural space, leading to a positive P_el(L) as P_alv = 0 (= atmospheric pressure). In contrast, P_el(L) ≤ 0 is associated with unphysiologic conditions resulting in lung collapse, such as open-chest or general anesthesia with mechanical ventilation.

The transpulmonary pressure (P_L) has been advanced as a surrogate of the elastic recoil pressure of the lung when the alveolar pressure (P_alv) can be approximated by the pressure at the airway opening (P_ao). This occurs when respiratory flows are zero (usually at end expiration and end inspiration) and no gas trapping exists¹³ :

Hence, a positive transpulmonary pressure throughout the respiratory cycle is required to maintain alveolar expansion.

General anesthesia, mechanical ventilation, and surgical interventions produce several biophysical factors promoting lung tissue collapse (fig. 2). Three major interrelated collapsing factors influence the balance of forces discussed above and contribute to perioperative atelectasis: increased pleural pressure, low alveolar pressure, and surfactant impairment. As a result of these factors, continuous or intermittent airway and alveolar closure occur, presumably more commonly the first than the latter.^14,15 

Pleural pressure is the pressure within the pleural cavity. It varies regionally across the pleural space depending on anatomical and physiologic interactions between the lung parenchyma, chest wall, and gravity.¹⁶  General anesthesia affects such interactions increasing regional pleural pressure (e.g., dorso-caudal in supine patients), resulting in negative transpulmonary pressure and compressive atelectasis (fig. 2).

The chest wall can be understood as composed by two functional portions: an elastic portion represented by the rib cage and abdominal wall and a constant weight component exerting a hydrostatic pressure represented by the abdomen. Changes in these portions will affect pleural pressure and lung expansion in the perioperative period, in line with the previously discussed equilibration of forces throughout the lung. The diaphragm is the primary muscle of lung ventilation and, consequently, significantly contributes to lung expansion and atelectasis development during anesthesia. For example, in anesthetized intubated patients without cardiopulmonary disease, phrenic nerve stimulation to produce diaphragm contraction reduces atelectatic area by approximately 33% as compared to mechanical ventilation with equal tidal volumes.¹⁷ 

In supine spontaneously breathing humans, diaphragmatic displacement and lung expansion are larger in dependent than nondependent lung regions.^18,19  This is due to the diaphragm dome shape with a smaller dependent radius of curvature, higher dependent stretch producing more favorable dorsal length-tension relationships, and possibly larger number of muscle fibers²⁰  and higher compliance of the crural than costal diaphragm.¹⁸  Diaphragm tension reduces the transmission of abdominal pressure to the lungs.²¹  Reduction or loss of such diaphragmatic tone during anesthesia thus affects the net balance of stresses acting on the lungs, not only reducing the preferential dependent displacement of the diaphragm but also facilitating the transmission of abdominal pressure to the lungs. This results in a cephalad shift of the dependent diaphragm with dependent lung compression and atelectasis^22,23  and no change or caudad shift of nondependent regions.^19,23  Relaxation of accessory respiratory muscles such as intercostals, scalenes, and sternocleidomastoids further contribute to reduction in cross-section chest area and lung aeration (fig. 3). In spontaneously breathing normal subjects receiving volatile anesthesia, the activity of parasternal muscles is abolished, and phasic expiratory activity in abdominal and lateral rib cage muscles is enhanced,²⁴  contributing to caudad-dependent atelectasis.²³  Muscle paralysis compounds to the dependent cephalad shift of the diaphragm and atelectasis during general anesthesia,²³  as the balance between alveolar pressure and the gravity-dependent hydrostatic pressure of abdominal contents becomes the main determinant of diaphragm motion.¹⁹  Ultimately, atelectasis that is preferentially dependent and caudad is detected in 90% of a broad population of patients without cardiopulmonary disease undergoing general anesthesia,²⁵  with up to 20 to 25% of initially normal lung either atelectatic or poorly aerated in transverse computed tomography during anesthesia.²⁶ 

Increased abdominal pressure as present with pneumoperitoneum, obesity, abdominal compartment syndrome, peritonitis, or abdominal shift of intrathoracic blood^27,28  produces further imbalance of net stresses on the lung, because it exposes the dorso-caudal lung to higher pleural pressure and susceptibility to atelectasis,^2,29  with cephalad shift of both diaphragm and intra-abdominal organs.^19,23,27  Of note, cephalad-outward movement of the lower ribs potentially produced by those factors can increase the cross-section of the lower chest and partially compensate for the loss of lung volume.²¹ 

Although gravity has been frequently cited as a key determinant of lung expansion, cephalocaudal gradients of lung expansion present in large animals³⁰  and humans^22,31  indicate the relevance of factors other than gravity. These factors include the matching of the lung to the thoracic cavity and the partially independent displacement of lobes, which are relevant determinants of regional lung expansion in supine and prone positions beyond gravitational factors.^30,32  Perioperative chest wall reshaping is influenced by body position, with proning allowing for recruitment of dorso-caudal lung.³³ 

Postoperative respiratory muscle dysfunction, particularly diaphragm dysfunction, has been documented after abdominal,³⁴  thoracic,³⁵  and cardiac surgery.³⁶  It facilitates the development of atelectasis as demonstrated by the significantly larger fraction of patients with atelectasis 24 h after thoracic surgery in the presence of postoperative ultrasound-diagnosed diaphragmatic dysfunction (35%) than in its absence (13%).³⁵  Diaphragm dysfunction can persist from a day to a week^37,38  and even up to a year.³⁹  It can occur because of direct injury to the diaphragm³⁶  or phrenic nerve⁴⁰  or because of indirect factors such as phrenic nerve dysfunction^37,41  and impaired thoracoabdominal mechanics.⁴²  These factors could compound with previous diaphragmatic compromise, for example, as present in neuromuscular disorders, sepsis,⁴³  abdominal hernias,⁴⁴  and potentially obesity.⁴⁵  Of note, diaphragmatic function could conversely affect regional lung inflammation by producing local high transpulmonary pressures as shown by the observation that spontaneously breathing lung-injured pigs exposed to low positive end-expiratory pressure (PEEP) present more dependent lung inflammation than those receiving high PEEP.⁴⁶ 

Although anesthetics (e.g., isoflurane, sevoflurane, and propofol) can compromise diaphragmatic function,^47–49  they do not affect contractility.⁵⁰  The diaphragmatic electromyographic activity can also be impaired by unwarranted administration of cholinesterase inhibitors, for example, neostigmine followed by sugammadex in humans⁵¹  or neostigmine administered after full recovery from neuromuscular block in rats.⁵² 

The weight of the nondependent lung compresses the dependent lung and pleural space determining pleural pressure increases along the vertical axis¹¹  with transpulmonary pressure reduction in dependent lung regions.^53,54  Pulmonary edema increases the weight of the lung tissues, increasing the risk of dependent atelectasis caused by the superimposed hydrostatic pressure.^54,55  Body position influences the effect of intrapulmonary compression by modulating the volume of dependent lung. For instance, the triangular shape of the lungs with the large dorsal base results in a greater volume of dependent lung in the supine than in the prone position.^55,56 

In supine patients, the mediastinal weight, particularly the heart, has been associated with pleural space compression and preferential retrocardiac lung collapse.⁵⁷  Pleural effusions may also compress the lung. However, the effusion volume does not entirely translate into compression⁵⁸  because of the compliance of the chest wall. For instance, in ARDS patients, chest wall expansion in the presence of a pleural effusion accommodates ~70% of the effusion volume.^58,59 

The concept of critical opening pressure has been introduced as the minimal alveolar pressure (P_alv in the previously described formulation) required to counteract the regional effect of collapsing forces.⁶⁰  Accordingly, lung units are expanded when alveolar pressure is higher than the critical opening pressure. A parallel concept is that of critical closing pressure, i.e., the alveolar pressure below which open lung units collapse. Mean closing pressure has been estimated as 6 cm H₂O in a small number of anesthetized mechanically ventilated patients,⁶¹  an interesting value to compare to the usual initial clinical setting of 5 cm H₂O. As determined experimentally in animal and computational models,^62,63  the critical closing pressure is lower than the critical opening pressure due to lung hysteresis, i.e., the difference between the inspiratory and expiratory components of the pulmonary pressure–volume curve, produced by opening of previously nonaerated peripheral airspaces.⁶⁴  Due to the vertical dependence of pleural pressures, critical opening pressures are higher in dependent regions as lung regions exposed to positive pleural pressure require alveolar pressures higher than these pleural pressures to achieve positive transpulmonary pressure and expansion.

The rationale for the use of PEEP derives from such a concept, ultimately aiming to keep alveolar pressures above critical closing pressures at end-exhalation to prevent lung collapse. Local variation in critical opening and closing pressures conditions the regions expanded and kept inflated throughout the breathing cycle at a given PEEP. Even normal lungs, when mechanically ventilated without PEEP for many hours, will progressively lose aeration preferentially in dorsal regions.⁶⁵  Of note, lung hysteresis imply that the PEEP required to keep lung regions open is lower than that required to open them (fig. 4).⁶⁶  This provides support to the practice of recruiting the lungs at pressures higher than those used during steady state mechanical ventilation.

The transmission of upper airway pressures to distal lung regions, i.e., the proximity between P_alv and P_ao, depends on the patency of regional airways (fig. 2). Increased airway resistance, secondary to airway constriction, obstruction, or compression¹³  determines a pressure drop in the distal lung with local alveolar pressures potentially lower than tracheal pressures (P_alv<P_ao). Obstructive atelectasis is the term used to describe lung collapse resulting from airway obstruction, usually caused by mucous plugs and retention of secretions superimposed or not to airway constriction. Anesthetics dose-dependently compromise ciliary motility of respiratory epithelial cells (isoflurane, ketamine, thiopental) potentially facilitating obstruction, and this compromise seems weaker with sevoflurane,⁶⁷  fentanyl or propofol.⁶⁸  Mucociliary clearance appears to be more compromised by cuffed endotracheal tubes than laryngeal mask airways.⁶⁹ 

Lung units presenting larger alveolar gas outflow than inflow will ultimately collapse. This is the basic concept of absorption atelectasis.⁷⁰  Such an imbalance is mostly determined by low local alveolar ventilation-to-perfusion (⁠$V.$_A/$Q.$⁠) ratios and high fraction of inspired oxygen. Low $V.$_A/$Q.$ determines low inflow of fresh air into alveoli in relation to the local perfusion associated with O₂ absorption. The lowest $V.$_A/$Q.$ ratios are found in the most gravity-dependent regions,⁷¹  where intraoperative atelectasis is typically present.^72,73  High fraction of inspired oxygen facilitates outflow as alveolar gas absorption by capillary blood is higher with O₂ than with gases presenting lower blood:gas solubility as nitrogen.^74–77  Lower mixed venous oxygen content further increases the rate of oxygen absorption (fig. 2).

Several factors during general anesthesia could impair surfactant function and contribute to the development of atelectasis: inflammatory response, mechanical ventilation, high oxygen concentration, anesthetics, and pulmonary edema. While these factors produce minimal compromise in short uncomplicated cases, they could become relevant as the surgical insult and patient compromise increase.

Inflammatory lung injury resulting from endotoxemia and bacteremia lead to reduced production of surfactant phospholipids,⁷⁸  increased surfactant turnover,⁷⁹  decreased tubular myelin⁸⁰  and altered surfactant protein A gene expression.⁸¹  Inflammation with surfactant dysfunction is also present during severe respiratory failure and pneumonia.⁸²  This dysfunction can be mediated by inflammatory cytokines,⁸³  proteases secreted from immune cells,⁸⁴  and increased proteins in alveoli.⁸⁵  In line with these findings, in vivo lung neutrophilic inflammation developing within few hours of mechanical ventilation in poorly aerated areas was reported in association with surfactant dysfunction,⁸⁶  with inflammation worsened by endotoxemia.⁸⁷  The translational relevance of such data is suggested by the quantitative and qualitative surfactant impairment observed in surgeries associated with large systemic and pulmonary inflammatory response (e.g., on-pump cardiac surgery) or lung injury.^88,89 

Mechanical ventilation impairs surfactant production by type II alveolar cells^90,91  through alveolar overstretching,⁹²  under-stretching⁹³  and monotonic stretching.⁹⁴  Conversely, deep breaths⁹⁵  and biologically variable ventilation⁹⁶  increase the release of active surfactant. Short-term exposure to high oxygen concentrations (100%) adversely affects surfactant function by increasing its susceptibility to rupture.⁹⁷  Suggestion that inhaled anesthetics could compromise surfactant biosynthesis in time- and dose-dependent patterns comes to date essentially from in vitro studies and is rapidly reversible after discontinuation.⁹⁸  Pulmonary edema, e.g., from fluid overload, can change surfactant activity and increase surface tension, potentially by the loss of surfactant into the edema fluid.⁹⁹  For permeability pulmonary edema, surfactant function can be inactivated by proteins in alveoli secondary to barrier disruption,^82,100 via increasing its conversion to non–surface-active forms.⁸⁵ 

Functional residual capacity is reduced in the supine position, and further by general anesthesia and muscle paralysis. Such decrease is associated with loss of muscle tone, and related reduction of the cross-sectional area of the thorax,²³  cephalad displacement of the diaphragm,¹⁹  increased curvature of the vertebral column,²⁴  and increased intrathoracic blood volume.²²  Of note, change in diaphragm displacement, thorax cross-section or functional residual capacity during general anesthesia were found not to correlate to the magnitude of atelectasis^22,23  suggesting the relevance of other factors for ultimate lung collapse, such as intrathoracic blood volume and decreased volume of the aerated lung.

Atelectasis is associated with lower respiratory system and lung compliances. This is because closure of alveoli and small airways corresponds to the low-volume portion of the pressure–volume curve associated with smaller change in lung volumes in relation to applied pressures, i.e., low compliance (fig. 4).^101,102  Indeed, in mechanically ventilated dogs, the respiratory system compliance is linearly related to lung volume.¹⁰³  Alveolar and airway closure is affected by supine position and may be further influenced by obesity (fig. 5).¹⁰⁴  Higher driving pressures (P_plateau – PEEP = tidal volume/respiratory system compliance) consequently also ensue.¹⁰⁵  End-expiratory lung volumes can be lower than closing capacity during mechanical ventilation likely due to loss of radial traction from parenchymal interdependence.^10,106  The same mechanism may contribute to the increased lung and respiratory system resistance associated with atelectasis in obese patients.¹⁰⁷  Of note, once atelectasis settles, the lung pressure–volume curve also becomes abnormal, further compounding the loss of lung volume.

The most clinically evident pathophysiologic effect of pulmonary atelectasis is hypoxemia.^1,102  Impaired blood oxygenation has been described during routine general anesthesia with both controlled¹⁰⁸  and spontaneous ventilation,¹⁰⁹  and correlated with the degree of atelectasis.^22,110  The mechanisms are low $V.$_A/$Q.$ ratios and intrapulmonary right-to-left shunt.^109,111  In experimental and clinical studies, alveolar recruitment promptly reverses gas exchange dysfunction.^1,101,112  Chronic obstructive pulmonary disease patients are less susceptible to oxygen absorption atelectasis¹¹³  and may develop less atelectasis during mechanical ventilation after cardiopulmonary bypass,¹¹⁴  presumably due to a combination of loss of elastic recoil and high airway resistance.

Hypoxic pulmonary vasoconstriction is a reflex constriction particularly of distal pulmonary arteries but also venules in response to hypoxia.^115,116  Oxygen sensing occurs at the alveolo-capillary level,¹¹⁷  with endothelial cell depolarization and retrograde propagation of the signal via gap junctions to upstream arterioles, where it is transmitted to pulmonary arterial smooth muscle cells to produce vasoconstriction.¹¹⁷  In health, the onset of the reflex to hypoxic gas mixtures occurs within seconds, and progresses into a plateau period (phase 1) of at least 20 min, and then a further rise (phase 2) starting after ~43 min and reaching a peak at ~2 h.¹¹⁸  This pattern is consistent with the response times to atelectasis observed in a dog model.¹¹⁹ 

Hypoxic pulmonary vasoconstriction mitigates right-to-left shunt diverting blood away from poorly oxygenated lung areas, thus supporting adequate arterial oxygenation during general anesthesia.^120,121  The more effective the hypoxic pulmonary vasoconstriction, the lower the effect of atelectasis on right-to-left shunt and consequent hypoxemia. In normal humans, 47% of blood flow is diverted from a hypoxemic lobe within 5 min.¹²¹  Factors that impair hypoxic pulmonary vasoconstriction accentuate hypoxemia produced by atelectasis: pneumonia,¹²²  ARDS,¹²³  endotoxemia,¹²⁴  sepsis,¹²⁵  calcium channel blockers,¹²⁶  atelectatic size,¹²⁷  or volatile agents (isoflurane, sevoflurane, desflurane) above 1 minimum alveolar concentration.^128–130  Mechanical factors associated with atelectasis could also impair hypoxic vasoconstriction; for example, unstable alveoli in the collapsed large animal lung could stent open pulmonary corner vessels through outward radial traction, reducing the hypoxic constriction.¹³¹  Intravenous anesthetics present a discrete to nonexistent effect on hypoxic vasoconstriction, because no blunting has been found with propofol,¹³²  ketamine,¹³³  or fentanyl,¹³⁴  whereas others found even oxygenation improvement with propofol.^133,135  During general anesthesia, the significant correlation between shunt and atelectatic area calculated from computed tomography in patients without significant lung disease indicates that hypoxic pulmonary vasoconstriction is not complete and contributes to hypoxemia even in normal lungs.¹¹⁰  In disease conditions, a strong correlation between shunt and atelectatic volume suggests more compromised hypoxic pulmonary vasoconstriction such as observed after mitral valve surgery when compared to coronary artery bypass graft surgery.¹³⁶ 

In normal humans, increases in the mean pulmonary artery pressure of ~10 mmHg are observed due to hypoxic vasoconstriction with arterial oxygen saturation changes from more than 95% to 85 to 90% and of ~16 mmHg with arterial oxygen saturation of 75 to 80%, corresponding to a duplication and more of the pulmonary vascular resistance.¹³⁷  In severe cases, right ventricular dysfunction can ensue as a consequence of the afterload elevation during atelectasis.^138,139  Accordingly, alveolar recruitment reduces pulmonary vascular resistance and increases pulmonary blood flow in patients after cardiopulmonary bypass¹⁴⁰  and in dogs with healthy or injured lungs,^141,142  subsequently improving right ventricular function.

Molecular (e.g., cytokine, chemokine, and pathways) and cellular (e.g., alveolar endothelial, epithelial, and inflammatory) changes during atelectasis imply significant local biologic responses and allow for insights into the mechanisms of lung injury. Atelectasis-associated lung injury could relate to direct or indirect, independent or synergistic local biologic effects such as inflammation, structural damage, hypoxic injury, surfactant dysfunction, and susceptibility to infection (fig. 6).

Links between atelectasis and inflammation have been indicated by clinical and experimental studies. Indeed, normal lungs exposed to 16 h of atelectasis and mechanical ventilation with low tidal volumes presented metabolic changes suggestive of incipient inflammation in large animals with aeration heterogeneity comparable to that of humans.⁶⁵  The hallmarks of atelectasis-related inflammatory response include the presence of cytokines and immune cells and their functional alteration (table 1).

Atelectasis is often associated with local production of inflammatory cytokines. Multiple clinical studies of one-lung ventilation^145,146,153  have reported increased levels of proinflammatory cytokines in the atelectatic lung, such as interleukins 1, 6, and 8 and tumor necrosis factor α, all involved in inflammatory injury. These cytokine levels were directly related to the duration of atelectasis^146,153  and potentially increase the susceptibility to postoperative pulmonary complications in patients undergoing lung resection surgery.¹⁴⁵  Moreover, atelectasis could result in significant concentrations of chemotactic cytokines, including chemokine (C-X-C motif) ligand 1,¹⁴⁴  a potent neutrophil chemoattractant; platelet-activating factor,¹⁴⁷  a mediator of platelet aggregation and degranulation, and leukocyte chemotaxis; and keratinocyte-derived chemokine,¹⁴³  another immune cell chemoattractant particularly for neutrophils. As a result, cytokines increased in atelectatic areas may cause direct injury and additionally act as homing molecules recruiting cells (e.g., neutrophils) into these regions that could further magnify damage, for example by releasing injurious cytokines.

Atelectasis contributes to inflammatory cell infiltration, at least in part through the inflammatory cytokines described above. For example, neutrophils, key immune cells in the inflammatory response and tissue damage are increased in bronchoalveolar lavage fluid of atelectatic regions in mechanically ventilated patients or spontaneously breathing dogs when compared to fluid obtained before atelectasis.^147,149,154  Those inflammatory infiltrates related to atelectasis duration¹⁵⁴  and were further magnified by systemic endotoxemia.¹⁵⁵  Additionally, atelectasis by itself can alter cellular immune function; for example, it can enhance alveolar macrophage cytokine secretion in rats,¹⁵¹  impair macrophage phagocytosis against bacteria in vitro in piglets,¹⁵²  and reduce local bronchoalveolar lymphocyte function in dogs.¹⁴⁹ 

Current evidence reinforces the concept that atelectasis produces an inflammatory response with pathophysiologic mechanisms different from those occurring in aerated lung regions. Indeed, different transcriptomic patterns in immunity have been documented recently in the atelectatic versus ventilated sheep lung, with fewer NF-κB–related genes in sterile lungs and more interferon-stimulated genes in the presence of systemic endotoxemia.¹⁴⁸  Such regional differences have also been found in a one-lung ventilation rat model showing increased myeloperoxidase, a neutrophil marker, in atelectasis and CCL2, a macrophage chemoattractant, in aerated lung regions.¹⁴⁴  Accordingly, findings reporting the similarity of inflammatory injury between atelectatic and ventilated lung either in humans^145,153,156  or in animals^144,150  may actually derive from different underlying cytokine and genomics responses in aerated and atelectatic areas. Understanding such regional responses to atelectasis could help to identify potential treatments beyond the usual ventilatory interventions. For example, nanoparticle delivery of micro-RNAs (i.e., miR-146a) mitigated mouse lung injury during mechanical ventilation.¹⁵⁷ 

Pulmonary structure disruption is a hallmark of lung injury. Immobility (lack of cyclic stretch) associated with atelectasis could contribute to structural damage potentially by disorganization of actin networks,¹⁵⁸  loss of adherens junction,¹⁵⁹  and impairment of barrier properties.^160,161  A recent sheep study also provided genomic support to these findings by revealing in initially healthy atelectatic lungs dysfunction of the lung tissue transcriptome related to structural components: endothelium, epithelium, and actin cytoskeleton.¹⁵⁴ 

Other factors potentially present during atelectasis such as inflammation and ischemia can also lead to structural dysfunction, such as impairment of sodium and chloride channels (i.e., ENaCα),¹⁶²  an ATP-dependent process involved in alveolar fluid clearance; and injury of the endothelial glycocalyx layer,¹⁶³  a critical component for lung barrier homeostasis. Re-expansion of atelectatic lung, a common process after one-lung ventilation, could be another contributing factor to structural damage caused by oxidative stress and inflammation.^164,165  Increased capillary transmural pressure from reversal of hypoxic pulmonary vasoconstriction after re-expansion and decreased pulmonary vascular resistance could mechanically injure the basement membrane of the alveolar–capillary lining.¹⁶⁶ 

Yes-associated protein 1 (YAP) signaling is a key pathway in the control of cell proliferation, apoptosis, and fate¹⁶⁷  and related to regulation of actin cytoskeleton dynamics¹⁶⁸  and alveolar epithelium repair and regeneration.¹⁶⁹  YAP signaling has been reported to be less activated in static pulmonary epithelial cells¹⁷⁰  and in atelectatic lung than in normally ventilated lung.¹⁵⁴  Together, these experimental studies suggest the potential role of YAP signaling and its possible use as a treatment target in structural dysfunction during atelectasis.

In line with those findings, microvascular endothelial disruption has been documented in atelectatic rat lung.¹³⁸  Such structural compromise could lead to lung edema,¹⁵⁶  microvascular protein leakage,^147,171  and even bacterial translocation to the bloodstream,¹⁷²  suggesting an additional mechanism contributing to increased lung permeability and decompartmentalization of the lung inflammatory response, increasing the risk for multiorgan dysfunction.¹⁷³  Potential therapies for structural dysfunction, such as β2 agonists directed at accelerating fluid clearance attempted in an acute lung injury trial, are examples of treatment targets derived from basic knowledge,¹⁷⁴  which illustrate the relevance of advancing the area.

Ultraprotective ventilation (limiting stress and strain with limited tidal volume and pressure) during extracorporeal membrane oxygenation (ECMO) for respiratory failure^175–177  is a recently discussed strategy in which the biologic effects of static and dynamic stretch may be relevant. The expected advantages of “lung rest” could benefit from barrier (e.g., epithelial and endothelial cells) protective effects of low cyclic stretch^178,179  and yet conflict with the damaging effects of lung immobility as well as the injurious effects of large static stretch on alveolar epithelial cells and extracellular matrix.^180,181  A recent trial reported no association of mechanical ventilation settings during the first 2 days of ECMO with survival of patients with severe lung injury.¹⁸²  A possible explanation is that the severity of the inflammatory response in these patients is so high that the ventilatory intervention would not be able to generate a biologic response. Thus, optimal ventilatory settings and length of their application for best lung recovery strategies in ECMO patients, including the best balance between immobility and cyclic load, remains an open question.

Lung collapse results in local hypoxia, a potent inducer of lung inflammation,¹⁸³  and microvascular injury.¹⁸⁴  Attenuating hypoxia or eliminating atelectasis by lung recruitment reverses lung injury induced by alveolar hypoxia.¹⁷¹  Experimental data suggest that such hypoxia-related lung injury may be associated with NF-κB–dependent chemokine (C-X-C motif) ligand 1 secretion from lung epithelial cells¹⁴⁶ ; macrophage recruitment and activation¹⁸³ ; decreased expression of lung neprilysin (a neutral endopeptidase)¹⁸⁴ ; and excess reactive oxygen and nitrogen species (superoxide anion radical O2•- and nitric oxide NO•).¹⁸⁵ 

The hypoxia activated transcription factor hypoxia-inducible factor 1 could be another important regulator in atelectatic tissue associated both with pro- and anti-inflammatory mechanisms.¹⁸⁶  Hypoxia-inducible factor 1α is increased and activated in nonventilated atelectatic rat lungs,¹⁴⁶  with distinct cellular effects. In myeloid cells, hypoxia-inducible factor 1α promotes acute inflammatory response through the regulation of glycolytic capacity.¹⁸⁷  In endothelial cells, hypoxia-inducible factor 1α reduces mitochondrial respiratory capacity and activates vascular inflammation by promoting glycolysis.^188,189  In contrast, in lung epithelial cells, hypoxia-inducible factor 1 contributes to anti-inflammation¹⁴⁶  and barrier protection.¹⁹⁰ 

Although surfactant dysfunction produces atelectasis as discussed above, conversely, atelectasis can lead to surfactant dysfunction. Classic studies reported that surfactant compression beyond 50% of its initial area, as potentially present during atelectasis, could result in film rupture on re-expansion and loss of function,^191,192  in line with increased surface forces associated with low PEEP (i.e., deaerated lung) during in vitro ventilation.¹⁹³  High surface forces in the alveoli causes transudation of proteinaceous fluid from capillaries into alveoli, further contributing to surfactant dysfunction.¹⁹⁴  Such dysfunction after atelectasis has been reinforced by clinical data from patients without cardiopulmonary disease, showing that surfactant phospholipids in bronchoalveolar lavage of atelectatic regions were lower after onset of atelectasis and remained low even after its resolution.¹⁴⁷ 

Pneumonia is a major postoperative pulmonary complication. Its incidence has been reported as 1.8% in American Society of Anesthesiologists status III patients undergoing noncardiothoracic predominantly abdominal and pelvic surgery,¹⁹⁵  3.5% after cardiac surgery,¹⁹⁶  and up to 25% after major lung resection.¹⁹⁷ 

Atelectasis has been often suggested as associated with pneumonia. The biologic compromise of the atelectatic lung immune defenses could facilitate the development of pneumonia, as detailed under “Inflammatory Response” above. Local depletion and dysfunction of surfactant secondary to significant atelectasis or after pulmonary edema could further compromise the anti-infectious response because surfactant possesses antimicrobial properties¹⁹⁸  and enhances macrophage phagocytosis and bacterial clearance.¹⁹⁹  Additionally, mucus plugging or impaired mucus clearance after long periods of atelectasis can increase the risk of infection by compromising mucociliary clearance against organisms entering the lung and trapping them within collapsed regions.^200,201 

Experimental studies support that atelectasis was associated with larger bacterial growth and pneumonia when bacteria were present or instilled into collapsed lungs.^172,200,201  Similarly, mechanical ventilation settings facilitating atelectasis (PEEP = 0 cm H₂O) increased lung bacterial burden in rabbits after tracheal bacterial instillation when compared with spontaneously breathing controls.²⁰²  Also, after the systemic intravenous injection of bacteria, the susceptibility to bacterial infection of atelectatic regions was greater than that of aerated regions.²⁰⁰  Of note, such effects of atelectasis might not be present when collapsed lung tissue is not exposed to any infectious agent, because atelectasis did not increase the incidence of pneumonia in dogs with noninfected lungs in a historical study.²⁰⁰ 

Clinical evidence has been more conflicting than such experimental studies. A large trial during major abdominal surgery indicated the high incidence of atelectasis and pneumonia in patients with ventilation settings predisposing to atelectasis (PEEP = 0 cm H₂O).³  However, the specific role of atelectasis or even an association could not be determined because large tidal volumes were combined with PEEP = 0 cm H₂O, and subsequent trials comparing high vs low PEEP in patients at different risk for alveolar collapse in similar settings did not show an effect of PEEP on postoperative pneumonia.^203,204  A potential explanation for those findings could be that the short-lasting intraoperative reduction of intraoperative atelectasis might not be enough for a longer-lasting effect in preventing infections through the first 5 to 7 days after surgery. Indeed, use of interventions addressing not only intraoperative but also immediate postoperative lung expansion resulted in less postoperative atelectasis and infectious complications in patients receiving lung expansion, suggesting a clinical effect.²⁰⁵  Such-hypothesis generating clinical results together with the basic science and translational findings suggest that interventions to at least minimize atelectasis could be relevant and require further investigation.

Lung tissues are continuously subjected to different mechanical forces associated with lung inflation during spontaneous and mechanical ventilation, as discussed above under “Physiologic Principles of Bronchiolar and Alveolar Expansion.” During atelectasis, lung mechanical forces contributing significantly to lung injury might be ascribed to different biomechanical processes, including cyclic opening and closing, stress concentration, and overdistension of the nonatelectatic lung (fig. 7).

Cyclic opening and closing of lung units (i.e., bronchioles and alveoli), presumably resulting from the unfavorable balance of forces acting on airways and alveoli, is a frequently cited but still incompletely understood mechanism for lung injury associated with atelectasis. Different processes potentially present during repeated opening and closing have been studied to explain the resulting injury, such as cyclic airway and/or alveolar reopening, the propagation of air–liquid interfaces with production of injurious longitudinal gradients of pressure, and shear stress.

Histological injury in lungs ventilated with PEEP below the inflection point of the pressure–volume curve (fig. 4), representative of PEEP insufficient to maintain lung units open,²⁰⁶  led to the concept of mechanical trauma caused by cyclic opening and closing. The critical opening pressure of an airway depends directly on airway fluid surface tension (γ) and inversely on airway radius (R; P_crit = 8.3γ/R).²⁰⁷  Such a relationship suggests a distribution of opening pressures along the airway tree with higher critical pressures and presumably injury from tidal recruitment at smaller airways.

Airway closure has been documented in vivo by computed tomography imaging in injured experimental lung models.^208,209  Cyclic opening and closing of airways results in bronchiolar injury as reported in animal models ventilated with zero²¹⁰  or negative end-expiratory pressure.²¹¹  In addition, repetitive alveolar collapse and expansion, directly visualized in surfactant-deactivated lung using in vivo microscope,²¹²  produces histologic injury with thickened alveolar walls, significant intra-alveolar edema, and numerous neutrophils.^212,213  Experiments document significant regional ventilation heterogeneity in poorly aerated regions in healthy lungs comparable with those of humans compatible with intermittent airway closure and reversible with PEEP.²¹⁴  The extent to which cyclic opening and closing occurs in human lungs and contributes to injury remains to be defined.

Surface forces can importantly contribute to epithelial injury during ventilation of atelectatic regions and associated opening–closing of airways and alveoli.^215,216  Secretions, surfactant dysfunction, and alveolar edema affect the fluid lining the airway and can lead to the formation of liquid plugs or liquid bridges in the airway lumen in association with lung collapse. Mechanical or spontaneous ventilation of such airways and underlying alveoli produces the propagation of these gas–liquid interfaces; that is, movement of the liquid plugs/bridges by the incoming air (fig. 7). The mechanical forces acting on the epithelium lining of airways and alveoli resulting from this movement have been advanced as a key mechanism of cell injury in mechanical ventilation of atelectatic regions.^215,216  The relevance of this mechanism has been experimentally supported in large tidal volume ventilation of normal rat lungs by observation of more wounded epithelial cells when gas–liquid interfaces were present (partial lung instillation of normal saline) than when lungs were either exclusively overdistended or completely saline-filled (i.e., no gas–liquid interface).²¹⁷ 

Cellular injury produced by propagation of gas–liquid interfaces include cell detachment and necrosis,²¹⁸  cell membrane fracture,²¹⁹  impairment of cell–cell adhesion,²²⁰  and deterioration of cytoskeletal structure.²²¹  Such damage increased with reduced airway compliance²²²  and diameter.²¹⁸  The rupture of liquid plugs can also lead to epithelial cell injury caused by fluid mechanical stresses in the vicinity and downstream of plug rupture²¹⁶  with associated inflammatory response.²²³  Different mechanical forces generated during interface propagation and acting on the epithelial cells lining opening airways include pressure and pressure gradients, shear stress, and shear stress gradients (fig. 7), with the pressure gradient likely being the primary determinant of mechanical damage.^215,224 

Surfactant treatment reduces the mechanical stress imparted by the propagation of gas–liquid interfaces.²²⁵  This is consistent with its success in neonatal use.²²⁶  In adults, failure of surfactant trials may have resulted from inadequate surfactant delivery to the distal airways and alveoli because of low instilled dose volume.²²⁷ 

Shear stress is defined as the force divided by the area parallel to the applied force (fig. 7). Cyclic opening and closing of the small airways or alveolar ducts could generate shear stress, acting on the collapsed and surrounding lung.^215,218  Although frequently mentioned as a common cause of injury during repeated opening and closing, no studies directly assessed shear stress in vivo. During the propagation of gas–liquid interfaces in vitro, shear stress at the air bubble cap was estimated as far greater than that in the regions upstream or downstream of the bubble tip.²¹⁵  However, theoretical investigations as detailed above suggested that shear stress is less important than the longitudinal pressure gradient in producing cell injury.^215,224 

Atelectasis-related mechanical injury can also be produced by “stress concentration,” first proposed by Mead et al.⁴  It is due to the distribution of mechanical forces in the three-dimensional lung structure around a region whose initially surrounding area is reduced by atelectasis. Stress concentration occurs in normal regions at the interface between open and closed lung, which are thus exposed to exaggerated stress (e.g., tethering stress described in physiology section) during ventilation (fig. 7).^228,229 

This mechanism could explain the injury observed around the atelectatic lung tissues, as presented in an ex vivo isolated, perfused rat lung model.²³⁰  Acting as a stress concentrator, atelectasis can generate structural alveolar injury and inflammation in the surrounding lung tissue.²³¹  Even microatelectasis can lead to histological epithelial injury caused by stress concentration as reported in a bleomycin-injured rat lung when ventilated with low PEEP and large tidal volume for 3 h.²³²  In addition, such stress concentration around atelectasis helps explain the clinical phenomenon of increased local neutrophilic activation at the interface between inflated and noninflated tissue in patients detected by positron emission tomography.²³³ 

Atelectasis leads to loss of aerated lung volume with redistribution of tidal lung volume during ventilation to the remaining smaller aerated lung (fig. 7).¹⁵¹  Thus, regional strain increases in such ventilated areas, with susceptibility to hyperinflation detectable by computed tomography.^234–236  This hyperinflation of aerated regions could promote higher lung inflammation than atelectasis at comparable low tidal volume and lower driving pressure.²³⁷  Consistent with these considerations, experimental findings in a rat lavage model of dependent atelectasis showed the coexistence of dependent atelectasis and remote nondependent lung injury characterized by distal airway injury and increased alveolar epithelial mRNA expression of inflammatory cytokines (e.g., interleukins 6 and 1 and macrophage inflammatory protein 2).²³⁸  Also, in initially healthy sheep with lung size and heterogeneity comparable to that of humans, receiving protective ventilation in the presence of mild systemic inflammation progressive atelectasis was associated with lung strain increased to areas of high aeration.¹⁵¹  Of note, these regions showed increased inflammation as assessed by positron emission tomography both in large animals¹⁵¹  and in patients with inflamed lungs,²³⁹  suggesting their contribution to ultimate clinical lung injury.

The perioperative period is associated with a profound imbalance of the physical forces that maintain, in the awake conditions, the physiologic expansion of the lung. Accordingly, pulmonary atelectasis, most frequently located in the dorso-caudal regions, represents an almost constant feature of general anesthesia. Hypoxemia and lowered respiratory system compliance are classical presentations of atelectasis at the bedside. Prolonged lung collapse and the associated biomechanical processes secondary to the ventilation of a heterogeneously aerated lung may actively participate in significant lung injury. The biologic response associated with atelectasis, before and after re-expansion, could further compound the injurious process. The impact of intraoperative pulmonary atelectasis on postoperative outcomes such as pneumonia and acute lung injury while presumed is still in need of high-level evidence. The presented information is expected to provide a basis for future inquire and physiologic-based clinical practice. Although the current focus on preventing postoperative pulmonary complications lies on using ventilator strategies to prevent atelectasis or overexpansion of atelectatic lungs, future approaches may take advantage of common or novel perioperative medications, which would address some of the sequalae of significant atelectasis at the cellular and molecular levels.

The figures were created with BioRender.com (Toronto, Ontario, Canada).

Supported by National Heart, Lung, and Blood Institute, National Institutes of Health (Bethesda, Maryland) grant Nos. R01 HL121228 and UH3 HL140177 (to Dr. Vidal Melo). Dr. Lagier was supported by research grants from Societe Francaise d’Anesthesie Reanimation (Paris, France), European Association of Cardiothoracic Anaesthesiology (Rome, Italy), and Fondation Monahan. Dr. Lee was supported by National Heart, Lung, and Blood Institute, National Institutes of Health grant Nos. R01 HL113022 and HL148781.

The authors declare no competing interests.