Вы находитесь на странице: 1из 11

144

Acute Respiratory Distress Syndrome:


Pathogenesis
Michael A. Matthay

I. PATHOPHYSIOLOGY OF PULMONARY EDEMA Role of Inflammation


IN ACUTE LUNG INJURY Role of Direct Toxicity
Vascular Fluid and Protein Exchange Biologic Markers
Increased Permeability Pulmonary Edema III. VENTILATOR-ASSOCIATED LUNG INJURY
Lung Physiology
Animal Studies
II. MECHANISMS OF ACUTE LUNG INJURY Clinical Studies
Path ological Findings IV. RESOLUTION OF LUNG INJURY
Mediators
Role of Infection V. CONCLUSIONS

This chapter focuses on the pathogenesis of acute lung injury where


(ALI) and the acute respiratory distress syndrome (ARDS).
Chapter 145 discusses clinical features and clinical manage- Jv = the net fluid-filtration rate (volume flow) across
ment. the microvascular barrier
Lp = the hydraulic conductivity (“permeability”) of the
microvascular barrier to fluid filtration (a mea-
PATHOPHYSIOLOGY OF PULMONARY sure of how easy it is for water to cross the barrier)
EDEMA IN ACUTE LUNG INJURY S = the surface area of the barrier
Pc = the pulmonary capillary (microvascular) hydro-
Pulmonary edema occurs when fluid is filtered into the lungs static pressure
faster than it can be removed. Accumulation of fluid may have Pi = the interstitial (“perimicrovascular”) hydrostatic
major consequences on lung function because efficient gas pressure
exchange cannot occur in fluid-filled alveoli. Lung structure πc = the capillary (microvascular) plasma colloid
relevant to edema formation and the forces governing fluid osmotic (or oncotic) pressure
and protein movement in the lungs has been the subject of πi = the interstitial (perimicrovascular) fluid osmotic
classic and more recent reviews and chapters, as noted in the pressure
“Suggested Reading” included at the end of this discussion. σd = the average osmotic reflection coefficient of the
barrier (a measure of the effectiveness of the bar-
Vascular Fluid and Protein Exchange rier in hindering the passage of solutes from one
side of the barrier to the other)
The essential factors that govern fluid exchange in the lungs
are expressed in the Starling equation for the microvascular
The Starling equation predicts the development of two differ-
barrier:
ent kinds of pulmonary edema. Increased pressure pulmonary
Jv = LpS[(Pc − Pi) − σd(πc − πi)] (1) edema occurs when the balance of the driving forces increases,

Copyright © 2008, 1998, 1988, 1980 by The McGraw-Hill Companies, Inc. Click here for terms of use.
2524
Part XVII Acute Respiratory Failure

forcing fluid across the barrier at a rate that can no longer be into the interstitial and alveolar spaces. The ability of the lym-
accommodated by lymphatic drainage. Increased permeability phatics to pump the excess filtrate away is increased when the
pulmonary edema occurs in the presence of ALI that damages lungs are injured. Maximal lung lymph flow increases more
the normal barriers to fluid filtration and allows increased flux when the microvascular wall has been injured than when hy-
of liquid and protein into the extravascular compartments of drostatic pressure alone is increased, but even this augmented
the lungs. lymphatic-pumping capability is taxed at lower driving pres-
Thus, pulmonary edema results from increases in ei- sures. If the epithelial barrier is injured, edema may accumu-
ther hydrostatic driving pressures (increased pressure edema) late readily in alveoli, because most of the resistance to fluid
or barrier conductance (increased permeability edema), or and protein flow into the alveoli is in the epithelial barrier.
both. What distinguishes the two types is barrier permeability, Increased permeability edema is often rapid in onset and pro-
which is normal in increased pressure edema, but abnormal gression because injured barriers offer much less resistance to
in increased permeability edema. Fluid flow into the lungs flow and because hydrostatic driving pressure is unopposed
is driven across the barrier in both types of edema by the by increases in osmotic pressure difference. Clinically, many
balance of pressures. ALI or ARDS results primarily from an patients with increased permeability edema have a low in-
increase in lung vascular permeability, although some cases travascular hydrostatic pressure, commonly measured as a
may be made worse by the presence of elevated lung vascular low or normal pulmonary capillary wedge pressure. In some
hydrostatic pressures. cases, this reflects the low intravascular pressures associated
with the underlying disease process, such as sepsis.

Increased Permeability Pulmonary Edema


Increased permeability pulmonary edema is caused by an in- Lung Physiology
crease in liquid and protein conductance across the barriers The effects of increased permeability edema on lung mechan-
in the lungs. The essential feature is that the integrity of the ics and gas exchange depend, in part, on the magnitude of
barrier to fluid and protein flow into the lung interstitium and edema accumulation. As with increased pressure edema, the
the alveoli is altered. Increased permeability edema is some- major effects on pulmonary mechanics occur with alveolar
times called noncardiogenic pulmonary edema, and the re- flooding. In experimental lung injury, functional residual ca-
sulting clinical syndromes in humans are commonly lumped pacity is decreased as a consequence of alveolar flooding; the
together as acute lung injury or the acute respiratory distress loss of units which can be ventilated accounts for most of the
syndrome. decrease in static lung compliance. Computed tomography
Accumulation of fluid and protein increases when the has provided new insights into structure-function relation-
lung endothelial and epithelial barriers are injured. If the rate ship in human ALI. In the early stage of lung injury, when
of fluid accumulation exceeds the rate at which it can be re- alveolar edema predominates, the lungs are characterized by
moved, increased permeability edema occurs. Because the a more homogeneous alteration of vascular permeability, and
barriers limiting fluid and protein flow into the lungs do not edema can accumulate evenly in all lung regions with a non-
function normally when the lungs are injured, the lungs are gravitational distribution.
not protected against edema by the usual safety factors. Al- Measurements of pulmonary mechanics in mechani-
though increases in fluid and protein filtration across the lung cally ventilated patients with ALIs show a decrease in static
endothelium can be removed by lymphatics and drained away lung compliance as a consequence of the loss of ventilated
from the alveolar walls as in increased pressure edema, much lung units. In addition, airflow resistance is increased as a re-
more fluid and protein are filtered at any given sum of driv- sult of decreased lung volume. Bronchospasm may add to the
ing pressures because the barriers to their flow are much less increase in airflow resistance and may be partially reversed
restrictive than normal. Edema formation in injured lungs is in some patients by administration of inhaled bronchodila-
very sensitive to hydrostatic driving pressures. Driving pres- tors. Chest wall compliance is reduced, probably because of
sures are often increased when the lungs are injured because of alterations in the intrinsic mechanical properties of the chest
the vasoconstrictive effects of inflammatory mediators such wall by abdominal distention, chest wall edema, and pleural
as thromboxanes, which may shift the main site of resistance effusion. Some investigators have reported differing respira-
to postcapillary venules, thus increasing hydrostatic pressure tory mechanics and response to positive end-expiratory pres-
at the microvascular fluid exchange sites, or because of effects sure (PEEP) during mechanical ventilation in patients with
on the heart as well as on the circulation. For example, ele- ARDS originating from pulmonary disease (e.g., pneumonia,
vated left atrial pressure, pulmonary venoconstriction, or an which causes consolidation) versus ARDS due to extrapul-
increase in cardiac output in sepsis can increase hydrostatic monary disease (which causes edema and subsequent alveolar
pressure at the microvascular fluid exchange sites. collapse).
Because the barriers are leaky, the protective osmotic Although the effects of surface forces on decreased lung
pressure differences across them are lost; driving pressure is compliance in ALI were once believed to be small, results of
unopposed by osmotic pressure, and even normal hydrostatic experiments in isolated rabbit lungs indicate that increased
pressure results in significant fluid and protein extravasation permeability edema may produce more severe mechanical
2525
Chapter 144 Acute Respiratory Distress Syndrome: Pathogenesis

changes than equivalent degrees of increased pressure edema. The injured alveolar epithelium is swollen, disorganized, dis-
In contrast, experiments in awake sheep have demonstrated continuous, and, frequently, detached from basement mem-
that similar degrees of pulmonary edema, regardless of mech- branes, which may be otherwise intact. The alveolar surface
anism, cause similar changes in compliance and gas exchange. may be covered by hyaline membranes. Type I cells are more
Other studies indicate that dynamic and static lung compli- severely damaged than type II cells. The thin cytoplasmic ex-
ances are decreased early in evolving lung injury. tensions of cells far from the nucleus, which cover the thin
Surfactant is strongly thromboplastic, and coagulation side of the alveolar-capillary barrier, may be most severely
may compound surfactant depletion when plasma proteins affected. The interstitium is widened by edema (especially
enter the airspaces. The injured lung may release substances in peribronchovascular cuffs) and may be filled with leuko-
that can interfere with the normal, low surface tension in the cytes, platelets, red blood cells, fibrin, and debris (especially
alveoli. In addition, activated neutrophils may impair surfac- near the alveolar walls). The microvascular endothelium is
tant function in vitro and degrade major surfactant apopro- relatively preserved; it usually shows little other than irregu-
teins through a combination of proteolysis and oxidant- lar, focal thickening due to cytoplasmic swelling or vacuoles
radical–mediated mechanisms. In studies using bronchoalve- and greater numbers of luminal leukocytes.
olar lavage, human lung surfactant obtained from patients at After about 5 to 10 days, the exudative phase is fol-
risk for ALI and from those with established ALIs has been lowed by a proliferative phase. The relative contributions of
reported to be abnormal in chemical composition and func- the original insult, repair processes, and effects of therapies
tional activity. Abnormalities may also be caused by inter- on this and subsequent phases are not well known. Some ab-
actions between surfactant and edema proteins, since plasma normalities occurring after the initial exudative phase appear
proteins (especially fibrin monomers, but also fibrinogen and to be related to effects of traditional modes of mechanical
albumin) interfere with surfactant function. Proteinaceous ventilation that used tidal volumes between 12 and 15 ml/kg
edema fluid has been associated with surfactant inhibition in predicted body weight.
several experimental models. Reabsorption of some of the edema fluid character-
Gas exchange is severely compromised in increased per- izes the proliferative phase. Fibrin may be prominent in alve-
meability edema because of both intrapulmonary shunting of oli and interstitium, and infiltration with inflammatory cells
blood and ventilation-perfusion inequalities. New evidence and fibroblasts, which may have been activated very early in
indicates that patients with early ALI have a marked increase the course of lung injury, may be seen. The alveolar epithe-
in pulmonary dead space fraction. This finding indicates that lium is often cuboidal and made up largely of proliferating
many ventilated lung units are not well perfused, although type II cells. The air-blood barrier may be thickened by in-
intrapulmonary shunting may also contribute to the elevated terstitial and epithelial enlargement. The pulmonary vascular
dead space. Not surprisingly, minute ventilation is typically bed may be partially or completely disrupted, and structural
twice normal (approximately 12 L/min) at the onset of ARDS. alterations may reduce its surface area.
Approximately 2 weeks after the initial insult, a final
stage may be observed in which fibrotic changes of the alveo-
lar ducts, alveoli, and interstitium predominate. Alveoli may
MECHANISMS OF ACUTE LUNG INJURY be obliterated, alveolar walls coalesced, and functional lung
units lost. The lungs may be emphysema-like, with extensive
This section focuses on the characteristic pulmonary bullous changes. Notably, even severe changes at any stage
pathological findings in patients with ALI and ARDS and may be reversible during a slow recovery back toward normal
the mechanisms responsible for ALI. lung function.

Pathological Findings Mediators


Based on several studies that included a preponderance of The most common clinical disorders associated with the
postmortem pathology, the light and electron microscopic development of ALI are pneumonia, sepsis, gastric aspi-
appearances of human and animal lung tissue in ALI have ration, and major trauma. Other, less common causes in-
been described. Exudative, proliferative, and fibrotic changes clude transfusion-associated lung injury, drug overdose, se-
usually appear in sequence. vere acute pancreatitis, and near drowning. The initiating
The earliest changes are marked by widespread alve- insult to the lungs occurs either via the airways or the blood-
olar and interstitial edema and hemorrhage. Hyaline mem- stream.
branes, composed of precipitated plasma proteins, fibrin, and The exact mechanisms by which the lungs are injured
necrotic debris are frequently found (Fig. 144-1). The alveo- have been the subject of intense investigation in humans, an-
lar epithelium may be more extensively damaged than is the imals, and cellular systems. Human studies have provided
vascular endothelium, even if the underlying insult is blood- descriptive data about the events that occur in the airspaces
borne. Widespread, local areas of destruction of type I alve- before and after the onset of lung injury. Studies using bron-
olar epithelial cells alternate with normal-appearing alveoli. choalveolar lavage or collection of pulmonary edema fluid
2526
Part XVII Acute Respiratory Failure

A B

Figure 144-1 A. A low-power light micrograph of lung biopsy specimen collected 2 days after onset of ALI/ARDS
secondary to gram-negative sepsis demonstrates key features of diffuse alveolar damage, including hyaline mem-
branes, inflammation, intra-alveolar red blood cells and neutrophils, and thickening of the alveolar-capillary mem-
brane. B . High-power view of a different field illustrates dense hyaline membrane and diffuse alveolar inflammation.
Polymorphonuclear leukocytes are imbedded in the proteinaceous hyaline membrane structure (black arrows). The
white arrow points to the edge of an adjacent alveolus, which contains myeloid leukocytes. (Histological sections in
A and B courtesy of Dr. K. Jones, University of California, San Francisco). C. Electron micrograph from a classic analysis
of ALI/ARDS showing injury to capillary endothelium and alveolar epithelium. LC = leukocyte within the capillary
lumen; EC = erythrocytes; EN = blebbing of the capillary endothelium; BM = exposed basement membrane where
the epithelium has been denuded; C = capillary; A = alveolar space. (From The ARDS Network: Ventilation with lower
tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome.
N Engl J Med 342:1301–1308, 2000, with permission.)

in patients following the onset of ALI have demonstrated a animal models do not completely reproduce all of the aspects
major acute inflammatory response beginning prior to clin- of ALI in humans, in part because human ALI evolves over a
ical recognition of ALI. The response peaks during the first longer period of time than can be studied in the laboratory.
1 to 3 days of clinically defined ALI and resolves slowly over In addition, the lungs of humans are exposed not only to the
7 to 14 days in patients who remain intubated. These stud- initial injurious insult, but also to the therapies that are used to
ies have shown the complexity of the evolving inflamma- treat ALI, such as mechanical ventilation. Experiments using
tory responses, characterized by accumulation of acute re- isolated cells have been helpful in testing specific concepts, but
sponse cytokines and their naturally occurring inhibitors, the complexity and redundancy of intact biologic systems is
oxidants, proteinases and antiproteinases, lipid mediators, not reproduced in simplified experimental systems. By design,
growth factors, and the collagen precursors involved in the most experimental work limits study to one causative agent,
repair process. thereby reducing actual clinical complexity to the simplicity of
Hypotheses regarding the mechanisms of lung injury a single experimental pathway. Increased permeability edema
have been tested in animal models and in vitro studies, and in humans is likely to be caused by interactions among a
several reviews have summarized the findings. The existing number of different pathways acting in parallel or series.
2527
Chapter 144 Acute Respiratory Distress Syndrome: Pathogenesis

Studies in isolated organs and small animals in which tory response mediated, at least in part, by neutrophils and
hemodynamic variables are not measured can be difficult to tumor necrosis factor (TNF). Endotoxin may also affect the
evaluate. Indices of lung injury, usually measured by the ap- clotting system and metabolic functions of the lungs, as well as
pearance of markers in lungs, lavage fluid, or perfusate, are predispose the lungs to development of pulmonary infections
not determined solely by the barrier function of the microvas- by increasing adherence of bacteria to injured endothelium.
culature. Indeed, when vascular endothelium is injured, fluid Exoproducts of bacteria, such as elastase and Pseudomonas
and protein movement from the vascular space into the lungs exoenzyme U, also have been shown to injure the lungs.
is sensitive to hydrostatic driving pressures and filtration sur- In addition to a direct role in the pathogenesis of lung
face area. Hence, the effects of experimental interventions injury, bacterial products may also have an indirect role by
may be caused by changes in these parameters and not by sensitizing the lungs to the effects of mechanical stretch.
changes in microvascular barrier function. Gram-negative lipopolysaccharide causes an acute inflam-
The effects of microvascular driving pressures and sur- matory response in the lungs of humans. Bacterial endotoxin
face area can be difficult to evaluate, even in large, instru- enhances the responses of human alveolar macrophages to
mented animals. In sheep and goats, interpretation of lung positive pressure ventilation; pretreatment of rats with in-
lymph fluid and protein flow changes are further compli- travenous endotoxin enhances cytokine production in the
cated by contributions of extrapulmonary lymphatics, physi- lungs during mechanical ventilation ex vivo. Furthermore,
cal forces acting on lymphatics, and possible intranodal mod- mechanical ventilation using moderate or large tidal volumes
ification of lymph. Data from experimental animal models increases the sensitivity of lung macrophages to endotoxin
suggest that at least two broad categories of mechanisms of in vitro and the expression of the endotoxin recognition
ALI are operative: (1) those that are indirect (i.e., require molecule, CD14, on lung cells in vivo. Endotoxin recognition
the participation of intermediary mechanisms, e.g., host de- pathways are increased in the lungs of patients with ARDS,
fenses); and (2) those that are direct (i.e., do not require in- and the biologic effects of endotoxin are amplified in the
termediary mechanisms; injury probably occurs as a result lungs of patients with lung injury. The synergism between
of contact between an offending substance and lung tissue). bacterial products and mechanical stretch suggests that in-
These categories overlap, since once the lungs are injured, terrupting these pathways might limit some forms of ALI in
inflammatory responses occur, which may compound the humans.
primary mechanism of injury. Three major hypotheses re- Increased permeability edema is associated with im-
garding the mechanism of ALI are discussed below. Although paired antibacterial defenses. In animal models, bacterial in-
discussed separately, they are interrelated. fections worsen ALI. The cause of impaired bacterial defenses
in acute ALI is not known. Bactericidal properties of the alve-
olar lining material might be altered in injured or flooded
Role of Infection lungs, and alterations in surfactant concentration and func-
ALI develops in 20 to 45 percent of patients with severe sepsis. tion may be important. Although neutrophils may be present
Increased microvascular permeability to albumin has been in large numbers in the bronchoalveolar lavage fluid of pa-
shown to accompany human sepsis, and infection and the tients with ALI, evidence indicates that the function of the
sepsis syndrome are major causes of ALI in humans. Patients neutrophils is compromised.
who develop shock in response to known or suspected in-
fection have a particularly high incidence of ALI, and the
mortality of patients with ALI associated with infection (i.e., Role of Inflammation
sepsis syndrome) is increased. ALI also appears to predispose Substantial evidence implicates host defenses and inflamma-
the lungs to infection, and delayed infection is an important tory responses in the underlying mechanism of many ALIs.
cause of morbidity in patients who survive the initial lung Neutrophils are a vital component of host defenses, and pa-
insult. tients with severe neutropenia are at increased risk of bacterial
The mechanism by which infection and sepsis syn- and fungal infections. On the other hand, neutrophils release
drome injure the lungs is not certain. The lung injury is likely toxic oxygen radicals, proteases, and other biologically active
related to factors other than direct damage by bacteria or mediators that initiate inflammation. Other important cells
other microorganisms, since the prognosis appears unrelated in pathogenesis include alveolar and pulmonary intravascular
to documented bacteremia or pneumonia. In experimen- macrophages, and eosinophils.
tal animals, intravenous infusions of live Pseudomonas aeru- Normally, the pulmonary circulation contains a very
ginosa or endotoxins from Escherichia coli or surgically in- large pool of marginated neutrophils that change shape in
duced peritonitis result in increased permeability pulmonary order to squeeze through the lung capillaries. When neu-
edema. Instillation of endotoxin into the airways of sheep trophils are activated, they stiffen and become less disten-
also leads to lung inflammation with variable degrees of lung sible. These neutrophils are retained for longer periods of
injury. P. aeruginosa produces lung injury in pigs, and E. coli time in the pulmonary microcirculation. Endothelial acti-
endotoxin administration injures the lungs of baboons and vation leads to increased expression of leukocyte adhesion
dogs; neutrophilic alveolitis is observed in rats and mice. ALI molecules, providing a second mechanism to slow the transit
caused by endotoxin in sheep is thought to be an inflamma- of neutrophils. Trapped neutrophils respond to chemotactic
2528
Part XVII Acute Respiratory Failure

gradients generated by chemokines produced by alveo- ing the idea that injury is caused directly by contact with the
lar macrophages and mesenchymal cells and migrate into respiratory epithelium in the airways and/or alveolar walls.
the airspaces. Activated neutrophils generate and release Inflammatory pathways are likely to be rapidly acti-
toxic substances (e.g., oxygen metabolites and granular con- vated following many types of direct lung injury, as proba-
stituents, such as proteases, and cationic lysosomal enzymes) bly occurs following aspiration of gastric secretions—one of
that disrupt the function of the microvascular and epithe- the most common clinical causes of ALI. Lung injury occurs
lial barriers. Normally, these barriers limit liquid and protein rapidly, especially to the epithelium. The injury is probably
flow out of the vascular space and into the alveolar spaces, related, in part, to the low pH of the aspirated stomach con-
mitigating development of permeability edema. tents (aspiration of gastric contents with pH greater than 2.5
Inflammatory responses also have the potential to in- is relatively benign; aspiration of gastric contents with pH
duce lung cell injury by activating cell death pathways, leading less than 2.5 causes severe pulmonary injury). Aspirated acid
to apoptosis. Bacterial products, such as Pseudomonas Exoen- is almost immediately neutralized. However, within hours,
zyme U and mechanical stretch, may lead to direct cellular proinflammatory mediators are released, the injured lung is
necrosis. Apoptosis is mediated by a family of death receptors, infiltrated with neutrophils, fibrin accumulates in the alveolar
including TNF and Fas receptors. The Fas ligand (FasL) is a 45 spaces, and further structural damage is seen on histological
kD peptide that is shed from the cell surface by the action of examination.
metalloproteinases. Biologically active soluble FasL (sFasL)
accumulates in the lungs of patients with ARDS, inducing
apoptotic death of human lung epithelial cells in vitro. Hu- Biologic Markers
man sFasL induces epithelial cell death in the lungs of rabbits; Considerable interest exists in finding a simple test of blood,
a monoclonal antibody that activates membrane Fas causes urine, or bronchoalveolar lavage fluid that would identify
alveolar wall apoptosis and fibrosis in the lungs of mice. patients at risk for, or in the earliest stages of, ALI, or that
Apoptosis and inflammation pathways intersect, as might predict clinical outcome.
stimulation of membrane Fas induces cytokine production Although products of complement activation have been
in human macrophages and inflammation in the lungs of proposed as markers, their serum levels correlate poorly with
rabbits and mice. In addition, lung injury may be able to lung injury. Measurement of circulating endotoxin is not ap-
trigger apoptosis pathways in distant organs, such as the kid- propriately sensitive or specific for the presence or risk of
ney, perhaps by increasing the concentrations of circulating developing lung injury. The same is true for measurements
sFasL. Thus, inflammatory responses may trigger cell death of release or activity of angiotensin-converting enzyme. Von
pathways, and cell death pathways triggered by sFasL may in- Willebrand factor (VWF) antigen may be useful as a plasma
duce inflammation in the lung alveolar environment. Recent marker of impending ALI in patients with nonpulmonary
human studies implicate apoptosis in human lung injury. sepsis. Recent work confirms that VWF levels are elevated in
the edema fluid and plasma of patients with ALI and cor-
relate with poor clinical outcomes. While increases in other
biochemical and inflammatory markers, including surfactant
Role of Direct Toxicity protein D and interleukin-6, correlate somewhat with lung
Inflammation is not required for all forms of ALI. ALI or injury and mortality, no simple biologic marker currently
ARDS can develop in neutropenic patients. A clinical trial serves in the same diagnostic capacity as do cardiac enzymes
using granulocyte colony-stimulating factor to increase the in evaluation of suspected acute myocardial infarction.
number and activation state of circulating neutrophils in pa- Because neutrophils are implicated in the mechanism
tients with severe pneumonia was not associated with an in- of many lung injuries, their detection in the lungs, assess-
creased incidence of ARDS. Lung injuries that do not require ment of their function, or assay of the toxic metabolites they
the participation of neutrophils have been described in animal release might be useful. For example, increased hydrogen per-
models. oxide levels have been measured in the breath and urine of
Direct lung injury is also thought to occur in humans. patients with ALI, presumably reflecting the presence of oxy-
Putative agents that directly injure the lungs include mechan- gen metabolites in the injured lungs. Evidence of increased
ical forces during mechanical ventilation, toxic and corro- oxidant activity has been reported in bronchoalveolar lavage
sive chemicals and gases (e.g., hydrochloric and other acids, fluid in patients with lung injury.
ozone, ammonia, chlorine, phosgene, nitrogen dioxide, the Finally, other mediators of inflammation in ALI have
vapors of cadmium and mercury, combustion products, and been studied. For example, increased levels of TNF are de-
oxygen, especially at high concentrations), ionizing radiation, tected in blood and bronchoalveolar lavage fluid in lung in-
aspiration of fresh water (near drowning) or hydrocarbon jury, but an association between TNF levels and development
compounds (e.g., kerosine, gasoline, and dry-cleaning fluid), of ARDS has not been found. Furthermore, elevated TNF
high temperatures (parenchymal lung burns from fires or ex- levels are found in patients with severe congestive heart fail-
plosions), and mechanical injuries (e.g., lung contusion from ure. Lipoxygenase products of arachidonic acid metabolism
nonpenetrating chest trauma or blast injury from explosions have been detected in pulmonary edema fluid, bronchoalve-
or lightning). Many of these injuries develop rapidly, support- olar lavage fluid, plasma, and urine, and elastase has been
2529
Chapter 144 Acute Respiratory Distress Syndrome: Pathogenesis

Figure 144-2 Multiple cellular responses and mediators contribute to alveolar-capillary membrane injury (right-
hand side) and the transition from normal alveolar structure and function (left-hand side) in the acute phase of
ALI/ARDS. Original investigations of the pathogenesis of ALI/ARDS searched for single mediators that provided final
common pathways to inflammation and alveolar edema. Current concepts of pathogenesis involve multiple molecu-
lar factors of several classes, a variety of responding cells, and an imbalance between injurious and reparative signals
and pathways. See text and Ware & Matthay (2000) and Matthay & Zimmerman (2005). (Reprinted from Matthay MA,
Zimmerman GA: Acute lung injury and the acute respiratory distress syndrome: Four decades of inquiry into pathogenesis
and rational management. Am J Respir Cell Mol Biol 33:319–327, 2005, with permission.)

detected in bronchoalveolar lavage fluid in the setting of lung Figure 144-2 depicts multiple pathways involved in the
injury. pathogenesis of ALI and ARDS in the context of normal
ALI follows a wide variety of insults of varying sever- and injured alveoli. Emphasis is placed on potential path-
ity. Furthermore, many abnormalities detected in ALI are ways for injury across the vascular endothelium and alveolar
found in other diverse, severe illnesses that do not involve the epithelium.
lungs. Therefore, the likelihood that any single marker that
unequivocally identifies the risk or the presence of ALI will
be found seems remote. An investigative focus on particular
subgroups of patients with common causes of injury, coupled VENTILATOR-ASSOCIATED LUNG INJURY
with study of much larger groups of more definitively diag-
nosed patients, might prove helpful. An approach that has The most important development of the last 10 years in our
not received much attention is investigation of the sensitivity understanding of the pathogenesis and treatment of ALI is
and specificity of combinations of biologic markers. The new recognition that the long-standing practice of mechanically
field of proteomics will expand this type of investigation and, ventilating patients with ALI or ARDS using high tidal vol-
perhaps, identify patterns of protein abnormalities that can umes and airway pressures actually worsens the injury. An-
be found in plasma, urine, edema fluid, and bronchoalveolar imal studies first suggested the potential contributory role
lavage in patients with ARDS. of high tidal volumes and elevated airway pressures in the
2530
Part XVII Acute Respiratory Failure

pathogenesis of lung injury; subsequently, clinical trials con- D levels. Additional clinical and experimental studies are un-
firmed the findings. derway. A reduction in lung endothelial and epithelial injury,
attenuated inflammatory responses, reduced edema forma-
tion, and more rapid resolution of lung edema are likely part
Animal Studies of the mechanism(s).
Animal experiments have shown that ventilation using high
tidal volumes may increase vascular filtration pressures; pro-
duce stress fractures of microvascular endothelium, alveolar
epithelium, and basement membranes; and cause lung rup- RESOLUTION OF ACUTE LUNG INJURY
ture (so-called ventilation-induced lung injury). The injury
appears to be due to increased lung excursions at high vol- In the last two decades, considerable progress has been made
umes (“volutrauma”), rather than the high-airway pressure, in understanding the mechanisms responsible for resolution
per se, since it can be prevented by limiting thoracic motion of lung edema. More limited progress has been made in un-
(e.g., by placing the chest in a cast). The concept of volu- derstanding the resolution of lung inflammation.
trauma was first established in 1974 when investigators found Considerable advances have been made in our under-
that modestly elevated tidal volumes, especially in the absence standing of the clearance of fluid and solute from alveoli.
of PEEP, caused lung edema in rats. Several years later, ad- Active sodium and chloride transport across the alveolar and
ditional animal studies further demonstrated the potential distal airway epithelial barriers into the interstitium drives
injurious role of high tidal volumes and elevated airway pres- edema fluid removal from the airspaces. The uninjured alve-
sures, an effect termed ventilator-induced lung injury (VILI). olar epithelium has a remarkable ability to rapidly clear fluid
Subsequent experiments demonstrated that VILI could also from the airspaces. Even when mild-to-moderate alveolar in-
induce release of several proinflammatory cytokines, injur- jury occurs, salt and water transport capacity is often pre-
ing the lung and other organs—a process referred to as served. In severe injury, when the barrier is disrupted, the
“biotrauma.” These animal studies stimulated clinical inves- capacity to clear edema is lost. The vascular endothelium be-
tigation that revolutionized the care of patients with ALI or comes the limiting barrier between the vascular space and
ARDS. airspace. Clinically, the capacity to remove some alveolar
edema fluid (as indicated by increase in edema fluid to plasma
protein concentration ratio) in the first 12 hours following
Clinical Studies ALI is a favorable prognostic finding; the associated mortal-
The compelling evidence from animal experiments and small ity rate is only 20 percent. In contrast, an inability to resorb
clinical trials prompted clinical studies aimed at testing the alveolar edema fluid early in the course of injury is associated
potential benefit of lower tidal volumes and reduced airway with a mortality of nearly 80 percent. Thus, the function of
pressures in management of ALI or ARDS. In a large, mul- the alveolar epithelial barrier early in the course of ALI may be
ticenter, National Heart Lung and Blood–sponsored trial of a useful prognostic index, serving as a marker of the severity
861 patients, mortality was reduced from 40 to 31 percent and extent of injury.
using a tidal volume of 6 ml/kg/ideal body weight and a lim- In uninjured, ex vivo human lungs, alveolar fluid clear-
ited plateau airway pressure of less than 30 cm H2 O. In this ance is increased by administration of salmeterol. In addition,
trial, use of small tidal volumes was associated with a lower experimental studies have shown that even in the presence of
incidence of nonpulmonary organ failure. The protocol for ALI and alveolar edema, alveolar fluid clearance can be in-
carrying out the lung protective ventilatory strategy is de- creased pharmacologically (e.g., by catecholamines), thereby
scribed in detail in Table 144-1. The results of the trial have representing a potential therapeutic intervention.
transformed the management of patients with ALI or ARDS. Clearance of protein from flooded alveoli is much
A follow-up clinical trial has shown that ventilation using slower (1 to 2 percent per hour) than clearance of fluid (10 to
the limited tidal volume and plateau pressure of the original 20 percent per hour), resulting in an increased concentration
study is associated with an overall reduction of mortality to of protein in airspaces. If the alveolar edema formed during
26 percent. In the follow-up study, although elevated levels increased lung vascular permeability clots, its removal from
of PEEP did not decrease mortality, the basic lung protective flooded alveoli may be slowed. Clotting may occur because
strategy was validated as effective. extravasation of plasma into airspaces can lead to clotting
The beneficial mechanism underlying the low tidal vol- system activation by surfactant or macrophage-derived pro-
ume strategy is unclear. An Italian study has shown that use coagulants.
of low tidal volumes in patients with ARDS attenuates the Most of the interstitial water in pulmonary edema lies
inflammatory response in both lungs and bloodstream, as in the peribronchovascular loose connective-tissue spaces,
measured by reductions in neutrophil and cytokine concen- rather than in alveolar walls. Because the lymphatic capillaries
trations in bronchoalveolar lavage and cytokines in circulat- are arranged to drain only the alveolar wall interstitium, this
ing blood. Other studies have confirmed a number of these route for edema removal is not significant for most interstitial
findings. In addition, a reduction in alveolar epithelial injury edema. A study in goats showed that lung lymph originates
appears likely, based on a decline in plasma surfactant protein mainly from alveolar wall interstitial fluid. The contribution
2531
Chapter 144 Acute Respiratory Distress Syndrome: Pathogenesis

Table 144-1
National Institute of Heart, Lung, and Blood, ARDS Network:
Lung Protective Ventilatory Strategy
Ventilator mode Volume assist-control

Tidal volume ≤ 6 ml/kg PBW

Plateau pressure ≤ 30 cmH2 O

Ventilation set rate, pH goal 6–35, adjusted to achieve arterial

pH ≥7.30, if possible

Inspiratory flow, I:E Adjust flow to achieve I:E = 1:1–1:3

Oxygenation goal 55 ≤Pao2 ≤80 mmHg or 88 ≤Spo2 ≤95%

FIO2 /PEEP Combinations

Fio2 0.3 0.4 0.4 0.5 0.5 0.6 0.7 0.7 0.7 0.8 0.9 0.9 0.9 1.0

PEEP, cmH2 O 5 5 8 8 10 10 10 12 14 14 14 16 18 18, 22, 24


(Further increases in PEEP to 34 cmH2 O allowed, but not required)

Weaning Attempts to wean by pressure support required when Fio2 /PEEP ≤0.40/8

PBW = predicted body weight


Male PBW = 50 + 2.3 [height (inches) − 60] or
50 + 0.91[height (cm) − 152.4]
Female PBW = 45.5 + 2.3[height (inches) − 60] or
45.5 + 0.91[height (cm) − 152.4]

I:E = ratio of inspiratory to expiratory duration


Pao2 = partial pressure of oxygen in arterial blood
Spo2 = oxyhemoglobin saturation measured by pulse oximetry
Source: Adapted from Acute Respiratory Distress Syndrome Network: Ventilation with low tidal volumes as compared with traditional tidal volumes for acute
lung injury and the acute respiratory distress syndrome. N Engl J Med 342:1301–1308, 2000, with permission.

of the lung lymphatic system to clearance of interstitial edema showed that the primary route of edema clearance is by vas-
in bronchoalveolar cuffs and interlobular septa is small. The cular resorption (60 percent of filtered water cleared over
maximum possible contribution by lung lymphatics to clear- 3 hours, including 42 percent by resorption into the blood-
ance of interstitial edema liquid is less than 10 percent, with stream and 18 percent by lymphatic, pleural, and mediastinal
airway loss of liquid through evaporation occurring at about drainage).
twice the rate of lymphatic clearance. Edema may also drain into the pleural space. Pleural
In a study of in situ perfused sheep lungs with experi- effusions are more common in increased pressure pulmonary
mentally induced low- and high-protein pulmonary edema, edema (25 to 50 percent of patients; usually on the right if
interstitial liquid was resorbed into the circulation in inverse unilateral). However, they occur in ALI as well (35 percent of
proportion to its protein concentration. Only a very small patients). As much as 25 to 30 percent of edema fluid may leave
fraction of interstitial edema was cleared by the lung lym- the lungs through the pleural space. A significant portion of
phatics during recovery from either type of edema. the interstitial edema probably follows the prevailing pressure
Some fluid from the loose peribronchovascular inter- gradient in the lungs to drain into the mediastinum, where it
stitium may drain directly into the bloodstream by crossing may be picked up by the lymphatics.
the walls of blood vessels in the lungs. A study of isolated Short-term alveolar protein clearance appears to pro-
sheep lungs made edematous by raising vascular pressures ceed primarily by paracellular diffusion. The process depends
2532
Part XVII Acute Respiratory Failure

on the size of the proteins. Most proteins are cleared in- A low tidal volume (6 ml/kg/ideal body weight), coupled with
tact, rather than as degraded, smaller fragments. However, a plateau pressure limit (less than 30 cm H2 O) has resulted in
a few specific proteins (e.g., vasoactive intestinal peptide and the first therapy demonstrated to reduce mortality in patients
gastrin) are degraded before being cleared. with ALI.
Receptor-mediated clearance may play a role. An Recently, the National Heart, Lung, and Blood Institute
albumin-binding protein (albondin) is expressed on lung mi- Acute Respiratory Distress Syndrome Network published the
crovascular endothelial cells. An antibody to this protein re- results of the Fluid and Catheter Treatment Trial (FACTT),
acts with cellular proteins of alveolar epithelial cells, which a large randomized trial comparing a liberal fluid manage-
also appear to have albondinlike binding sites for albumin. ment strategy to a conservative fluid management strategy in
In addition, a polymeric immunoglobulin receptor has been patients with ALI. Patients were also randomized to receive
described. The significance of these receptors in protein clear- either a pulmonary arterial catheter or a central venous
ance is, however, unclear. catheter for monitoring and fluid management. There were
Finally, the general consensus is that transcytosis (trans- no differences in clinical outcomes between the pulmonary
port via vesicles) is not a major mechanism for clearing bulk or central venous catheter utilization. In contrast, there was
quantities of albumin or other proteins from the alveolar a marked difference in outcome between the liberal and
space. Over the long term, phagocytosis and catabolism by the conservative fluid management arms of the study. Pa-
macrophages account for most protein clearance from the tients in the conservative fluid management arm had 2.5
alveolar spaces. All insoluble, precipitated proteins are re- more ventilator-free days than those in the liberal fluid man-
moved in this way. Macrophages are also ultimately respon- agement arm with concordant improvements in pulmonary
sible for removing senescent and dead neutrophils and other physiology. There was also a 2.9 percent reduction in the 60-
debris. The presence of a small, ciliated surface area of the day mortality rate in the conservative fluid management arm
distal airspaces suggests that the mucociliary route accounts compared with the liberal fluid management arm, although
for only a minor fraction of alveolar protein clearance. Com- this difference did not reach statistical significance.
plete clearance of alveolar protein from pulmonary edema by New insights into the pathogenesis of ALI suggest that
any route is slow. other therapies may also prove to be efficacious in reducing
Little is known about the mechanisms and signals that mortality in this common form of severe acute respiratory
regulate endothelial barrier function or how increased en- failure. A major development has been the ability to con-
dothelial permeability is returned to normal. duct large, prospective, randomized, clinical trials (e.g., those
sponsored by the National Heart, Lung, and Blood Institute)
to test a variety of therapies important in supportive patient
care, including use of mechanical ventilation, intravenous flu-
ids, and a variety of pharmacologic agents. Such trials have led
CONCLUSIONS to a better understanding of the pathogenesis of human ALI
and have confirmed the importance of ventilator-associated
Among the major advances in respiratory medicine and phys- lung injury. In addition, ancillary pathogenetic studies car-
iology over the last three decades has been the acquisition of ried out on biologic samples from clinical trials have advanced
important new knowledge on the physiology of fluid, so- our understanding of underlying mechanisms. In the future,
lute, and protein transport in healthy and diseased lungs. additional human studies will be needed to explore new ther-
Pulmonary edema, defined as the abnormal accumulation apeutic strategies suggested by basic investigation conducted
of extravascular lung fluid, is a pathological state that oc- in animal models of ALI.
curs when fluid is filtered into the lungs faster than it can be
removed.
The many causes of pulmonary edema are grouped into
two main pathophysiological categories: (1) increased pres- SUGGESTED READING
sure edema, which results from an increase in hydrostatic or
osmotic forces (or both) that act across the barriers that nor- Albertine K, Soulier MF, Wang Z, et al: Fas and Fas ligand are
mally restrict movement of fluid and solutes in the lungs; and up-regulated in pulmonary edema fluid and lung tissue of
(2) increased permeability edema, which is seen in ALI in patients with acute lung injury and the acute respiratory
which a breakdown of the normal barrier properties of lung distress syndrome. Am J Pathol 161:1783–1796, 2002.
endothelium or epithelium develops. Although these two dif- The ARDS Network: A trial of mechanical ventilation
ferent types of pulmonary edema share many features, usu- with higher versus lower positive end-expiratory pres-
ally they can be distinguished by careful clinical, radiological, sures in patients with acute lung injury and acute res-
and physiological evaluation. They also differ in treatment piratory distress syndrome. N Engl J Med 351:327–336,
and prognosis. 2004.
Major advances in the treatment of ALI have occurred The ARDS Network: Comparison of two fluid-management
because of the successful application of lung protective venti- strategies in acute lung injury. N Engl J Med 354:2564–
latory strategies early in the course of the illness (Table 144-1). 2575, 2006.
2533
Chapter 144 Acute Respiratory Distress Syndrome: Pathogenesis

The ARDS Network: Ventilation with lower tidal volumes as Matthay MA, Zimmerman GA. Acute lung injury and the
compared with traditional tidal volumes for acute lung acute respiratory distress syndrome: Four decades of in-
injury and the acute respiratory distress syndrome. N Engl quiry into pathogenesis and rational management. Am J
J Med 342:1301–1308, 2000. Respir Cell Mol Biol 33:319–327, 2005.
Dos Santos CC, Slutsky AS: Invited review: Mechanisms Matute-Bello G, Winn RK, Jonas M, et al: Fas (CD95) induces
of ventilator-induced lung injury: A perspective. J Appl alveolar epithelial cell apoptosis in vivo: Implications for
Physiol 89:1645–1655, 2000. acute pulmonary inflammation. Am J Pathol 158:153–161,
Dreyfuss D, Saumon G: Ventilator-induced lung injury: 2001.
Lessons from experimental studies. Am J Respir Crit Care Nelson S, Belknap SM, Carlson RW, et al: A randomized
Med 157:294–323, 1998. controlled trial of filgrastim as an adjunct to antibiotics
Eisner M, Parsons P, Matthay MA, et al and The ARDS Net- for treatment of hospitalized patients with community-
work: Plasma surfactant protein levels and clinical out- acquired pneumonia. J Infect Dis 178:1075–1080, 1998.
comes in patients with acute lung injury. Thorax 58:983– Nuckton TJ, Alonso JA, Kallet RH, et al: Pulmonary dead-
988, 2003. space fraction as a risk factor for death in the acute res-
Hastings R, Folkesson HG, Matthay MA: Mechanisms of alve- piratory distress syndrome. N Engl J Med 346:1281–1286,
olar protein clearance in the intact lung. Am J Physiol Lung 2002.
Cell Mol Physiol 286:L679–L689, 2004. Ranieri VM, Suter PM, Tortorella C, et al: Effect of mechani-
Hirsch J, Hansen KC, Burlingame AL, et al: Proteomics: cal ventilation on inflammatory mediators in patients with
Current techniques and potential applications to lung acute respiratory distress syndrome: A randomized con-
disease. Am J Physiol Lung Cell Mol Physiol 287:L1–23, trolled trial. JAMA 282:54–61, 1999.
2004. Robbins I, Newman JH, Brigham KL: Increased-permeability
Imai Y, Parodo J, Kajikawa O, et al: Injurious mechanical ven- pulmonary edema from sepsis/endotoxin, in Matthay M,
tilation and end-organ epithelial cell apoptosis and organ Ingbar, DH (eds), Pulmonary Edema. New York, Marcel
dysfunction in an experimental model of acute respiratory Dekker, 1998, pp 203–245.
distress syndrome. JAMA 289:2104–2112, 2003. Sakuma T, Folkesson HG, Suzuki S, et al: Beta-adrenergic ag-
Lum H, Malik AB: Regulation of vascular endothelial barrier onist stimulated alveolar fluid clearance in ex vivo human
function. Am J Physiol 267:L223–L241, 1994. and rat lungs. Am J Respir Crit Care Med 155:506–512,
Matthay MA, Folkesson HG, Clerici C: Lung epithelial fluid 1997.
transport and the resolution of pulmonary edema. Physiol Ware LB, Matthay MA: Alveolar fluid clearance is impaired
Rev 82:569–600, 2002. in the majority of patients with acute lung injury and the
Matthay MA, Martin TR: Pulmonary edema and acute lung acute respiratory distress syndrome. Am J Respir Crit Care
injury, in Murray JF, Nadel JA (eds), Textbook of Respiratory Med 163:1376–1383, 2001.
Medicine, 4th ed, Vol 1. Philadelphia, Elsevier Saunders, Ware LB, Matthay MA: The acute respiratory distress syn-
2005, pp 322–329. drome. N Engl J Med 342:1334–1349, 2000.