Mechanisms controlling the volume of pleural fluid and extravascular lung water

Steady state pleural fluid dynamics

Pleural fluid turnover averages 0.2 mL·kg⁻¹·h⁻¹ [8], which assures complete fluid renewal in ∼1 h. Fluid filters through the parietal pleura while drainage is mostly provided [7, 8] by the lymphatic stomata of the parietal pleura, which are opening 0.3–40 μm in diameter (either single or in cluster) and directly connecting the pleural space with the parietal lymphatic draining system [10]. The extension of the lymphatic network is mostly developed on the diaphragmatic surface where the density of stomata reaches the considerable value of 8,000·cm⁻², and on the mediastinal side where the density of stomata is so high that one speaks of a “cribriform membrane”; the mesothelial openings connect to an extended network of submesothelial lacunae [11]. In physiological conditions, lymphatic flow amounts to 3 μL·min⁻¹·mm⁻², flow velocity in initial lymphatics is of the order of 2 mm·min⁻¹ [1]. Under physiological conditions, a minor share of pleural fluid (<15%) could be drained through the visceral pleura in animals where the visceral pleura is perfused by pulmonary circulation. In this case there is a potential absorption gradient as the hydraulic pressure in pulmonary capillaries (∼10 cmH₂O) is lower than the plasma colloid osmotic pressure (∼25 cmH₂O) [1]. This mechanism is excluded in humans where the visceral pleura is perfused by the systemic circulation [12]. Some data suggest that active water transport operated by mesothelial cells plays a potential role in pleural fluid turnover [13].

The lymphatic conductance (Kl) is >10 times higher [7] than the filtration coefficient of the parietal pleura (Kf). In fact, the latter includes two membranes in series, the capillary endothelium and the parietal mesothelium, which create a considerable high-resistance system for water and plasma proteins [14, 15]. The visceral pleura are essentially excluded from pleural fluid turnover in physiological conditions because, due to its greater thickness, its permeability is ≥10-fold lower compared to that of the parietal pleura [14, 15].

There is a polarisation of filtration/drainage for pleural fluid due to intrapleural Pliq gradients as shown in figure 4⇓. Filtration mostly occurs in the less dependent portions of the cavity while drainage mostly occurs on the diaphragmatic and mediastinal surfaces [1] (as indicated by following radiolabelled markers). Some recirculation of pleural fluid has also been documented [16, 17]. Pleural liquid volume in physiological conditions averages ∼0.2 mL·kg⁻¹, only 1% higher than the theoretical minimum corresponding to the case where Pliq  =  Pabs [7]. Pleural fluid protein concentration averages ∼1.5 g·dL⁻¹ so that the physiological fluid/plasma protein concentration ratio (defined above) averages 0.2.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Polarisation of filtration/drainage processes in the pleural cavity and of intrapleural fluxes. White arrow: filtration; black arrow: lymphatic drainage; grey arrow: intrapleural fluxes.

Control of pleural fluid volume

It is certainly difficult to exactly differentiate between the three mechanisms potentially providing pleural fluid transports, namely lymph flow, active water reabsorption and Starling dependent absorption through the visceral pleura. However, it should be stressed that lymphatics, unlike other pathways for fluid removal, act as an efficient negative feedback system to regulate pleural fluid dynamics [7] through two mechanisms: 1) they set a rather sub-atmospheric pressure thanks to their powerful draining action; and 2) they can markedly increase draining flow in response to increased filtration [7, 8].

Figure 5⇓ presents a sensitivity analysis depicting the features of the control mechanism of pleural liquid volume based on lymphatic control [7]. The chequered area represents the extent of the control mechanism on pleural liquid volume (the controlled variable), tested by either increasing pleural fluid filtration rate or decreasing the maximum lymphatic draining flow. Point A corresponds to the physiological condition. Point B shows how, for a 10-times increase in filtration rate (due, for example, to an increase in hydraulic permeability), volume of the liquid (Vliq) would not even be doubled due to the increase in lymphatic draining flow. Point C shows that reducing the maximum lymph flow down to 1:10 of normal (e.g. for mediastinal lymphatic compression or mesothelioma) would still maintain Vliq close to normal, although it would be impossible to match an increase in filtration rate.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Sensitivity analysis describing the features of the lymphatic control on pleural liquid volume. Point A corresponds to the physiological condition. The ordinate shows that pleural liquid volume would only be slightly increased when microvascular filtration is increased 10 times (point B) or when the maximum lymphatic flow is decreased to 1:10 of normal (point C). Outside the chequered area (point D) no lymphatic control is operating.

The features of the control mechanisms have been described in an experimental animal model [1, 7]; thus, it appears reasonable to question whether the same mechanism operates in humans. The draining role of parietal lymphatics in humans is demonstrated by the epidemiological finding of the “black spots”, namely the accumulation of asbestos fibres around the parietal pleural lymphatic stomata [18]. The interpretation of this finding is that intrapleural translocation of ultrafine asbestos fibres reaching the pleural space [19] is flow-driven towards the pleural fluid draining sites; furthermore, the accumulation at the lymphatic entrance is due to the geometric hindrance of anisodiametric asbestos particles.

Pathophysiology of pleural effusion

When filtration overwhelms fluid drainage, pleural effusion results; point D in figure 5⇑, falling outside the chequered area, refers to a condition where the control mechanism becomes ineffective because the maximum lymphatic flow cannot match the increase in filtration rate. In this condition pleural effusion will ensue and the increase in pleural fluid volume will purely reflect a new equilibrium between the pressures acting across compartments.

Pleural effusion is favoured by an increase in systemic capillary pressure and an increase in permeability of the parietal/visceral pleura, a strong limitation to draining lymphatic flow. One shall consider that permeability × capillary surface area (Kf in equation 1) is a multiplicative factor and may increase by one to two orders of magnitudes to account for the increase in filtration rate.

Pleural effusion is classified as exudates when the fluid/serum total protein ratio (TPR) exceeds 0.5 [20–22]. However, this distinction may just result from the definition of a cut-off value. In fact, by plotting the average TPR values from a database of pleural effusions classified as either exudates or transudates [20] versus the corresponding reflection coefficient values (derived as ), an obvious inverse correlation is found (fig. 6⇓), meaning that the increase in TPR is the result of progressively more severe damage to the mesothelium that allows more plasma protein (mostly albumin, having a molecular weight of ∼70kDa) to leak. In transudates, σ is >0.5 while in exudates σ is <0.5. The expected experimental variability around the relationship (as suggested by the grey area in fig. 6⇓) reflects the complex interaction of the increase in mesothelial water permeability (either diffused or localised) and the decrease in protein reflection coefficient (also either diffused or localised). However, it appears justified to relate the degree of mesothelial lesion to the pleural fluid absolute concentration of lactic dehydrogenase [20] (indicated as free lactate dehydrogenase (FLDH)). A cut-off for FLDH has been proposed at 163–200 U·L⁻¹ for diagnostic separation between transudates and exudates [20, 23] and that would then correspond to σ ∼0.5.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Pleural fluid/serum total protein ratio (TPR) (corresponding to ) plotted versus protein reflection coefficient. The original data were taken from the two regressions presented in [20] referring to exudates and transudates. The figure intends to show that since TPR is a unique function of protein reflection coefficient, moving from transudates (·········) to exudates (–––) just corresponds to progressively increasing damage of the mesothelium.

It is now of interest to interpret some clinical observations concerning the occurrence of pleural effusion of various origin. We will do so keeping in mind that, due to the low permeability of the visceral pleura [1], the pleural cavity and the lung extravascular space appear to be rather independent compartments in terms of fluid exchanges in physiological conditions.

There is epidemiological evidence that pleural effusion is less frequent in right rather than in left heart failure [24, 25]. These observations are not framed into a specific set of functional parameters (microvascular permeability, lymph flow rate, etc.), therefore, only general considerations can be advanced. The increase in right atrial pressure and central venous pressure should entrain a corresponding increase in filtration rate at capillary level from the parietal pleura. Yet, if pleural effusion is rarely detected even when right atrial pressure exceeds 20 mmHg [25], the interpretation is that the corresponding increase in filtration rate of the parietal pleura can be compensated by lymphatic drainage (fig. 5⇑). The fact that pleural effusion is more frequent in left heart failure (it is present in ∼15% of class IV patients) should be regarded as the result of a marked increase in permeability of the visceral pleura; thus, in this condition pleural cavity and lung extravascular space are no more functionally independent (as they are in the physiological condition). The critical importance of increased permeability in causing pleural effusion is shared by Kroegel and Antony [26]. The combined action of hypoxia and inflammatory state can contribute to the increase in permeability. Furthermore, interstitial lung oedema, by generating positive interstitial pressure, may favour fluid filtration into the pleural cavity. Two more factors may be considered on comparing the visceral with the parietal pleura: the former has an incredibly extended underlying capillary network (2,000 cm²·g⁻¹ tissue) compared with the endothoracic fascia and has a ∼50% greater surface area due to the interlobar fissures. One shall note that pleural effusion may represent a “physiological finding” during lung water clearance in the first post-natal hours of term newborns [27].

Recovery from pleural effusion

The full recovery from pleural effusion is a long process mainly due to the fact that the lymphatic mechanism, although being quite adaptive (flow rate can increase by 20 to 30 times), might not match the potential increase in filtration rate resulting from the damage of the mesothelium. Data indicate that recovery ranges from weeks (for post-myocardial infarction and post-coronary artery by pass) to months (for tuberculosis and asbestosis) [28]. A decrease in pleural fluid volume can only occur when recovery from inflammation brings the filtration/drainage ratio below unity. This is critically dependent, in turn, upon a restoring a low mesothelial (visceral/parietal) permeability. The Starling mechanism could provide, in principle, fluid reabsorption in the initial phase, although the pressure gradient to sustain it remains quite poor because of the high compliance of the pleural space. Lymphatics will progressively contribute to liquid drainage at a later phase when the Starling dependent absorption is waning on progressively decreasing pleural liquid pressure.

Steady state fluid dynamics in the lung interstitial space

The diffusional fluxes of oxygen across the capillary wall are of the order of 10⁻² mL·cm⁻² per day, while capillary filtration is as low as 4×10⁻⁸ mL·cm⁻² within the same time, i.e. six orders of magnitude lower. Both fluxes occur through an incredible development of the capillary network (∼2,000 cm²·g⁻¹ lung tissue). The thinness of the air–blood barrier reflects a functionally “dry” condition, therefore, as much as for the pleural liquid. In addition, for the extravascular space of the lung one can speak of a “minimum” volume of water. In fact, when fluid fluxes increase due to alteration in fluid dynamics, this results in an impairment of diffusion.

The net filtration pressure gradient across the capillary endothelium is largely dominated by the rather sub-atmospheric interstitial pressure that reflects the equilibrium between the capillary filtration through the low-permeability endothelial barrier and the strong draining action of lymphatics. This sub-atmospheric pressure was measured by developing a technique to allow measurements in lungs normally expanded in the chest wall [3], thus maintaining the force vectors acting on the extravascular fluid as shown in figure 2⇑. An increase in elastic forces (over inflation) is a potent factor rendering Pi more sub-atmospheric [31]; a similar result would be caused by a decrease in alveolar surface tension favouring lung collapse.

Control of lung interstitial water

The lung is strongly resistant to the development of severe oedema. In fact, several mechanisms cooperate to control the volume of the extravascular fluid and to allow minimal variations relative to the steady state value when conditions of increased microvascular filtration are set, a crucial point to guarantee gas diffusion. A key role in buffering the increase in extravascular water is played by proteoglycans in two ways. First, their glycosaminoglycan chains can bind excess water in the interstitial space to form gel-like structures; in turn, the hydration level increases the steric hindrance of proteoglycans and decreases permeability, particularly at the level of the basement membrane. The other important role played by proteoglycans reflects their macromolecular assembly that assures low tissue compliance. In fact, as shown in figure 7⇓, in response to increased filtration, a minor increase in extravascular water (<10%) causes an increase in Pi by ∼15 cmH₂O (from approximately -10 to 5 cmH₂O) [9]. The significance of this mechanical feature is clear: the marked increase in Pi buffers further filtration (so called “tissue safety factor”) and, at the same time, it increases lymphatic flow (equations 2 and 4). Fluid accumulation occurs mainly in the thick portion of the air–blood barrier where the proteoglycan mesh is not as tight [29], thus avoiding thickening of the thin portion. Another potential factor contributing to control of lung water is the marked pre-capillary arteriolar resistance that contributes to keeping the capillary pressure at the physiological level of 10 cmH₂O [32, 33].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

The continuous line shows the time course of lung interstitial pressure when interstitial oedema develops (○). Note the marked increase in interstitial pressure for a minor change in extravascular water reflecting the very low compliance of the lung interstitial matrix. The dotted line shows the change in interstitial pressure when severe oedema develops (grey arrow). The decrease in pressure reflects the loss of integrity of the macromolecular structure of the extracellular matrix due to fragmentation of proteoglycans, which results in an increase in tissue compliance and in microvascular permeability. Restoring the filtration gradient leads to unopposed filtration and severe oedema (arrow). •: control.

Pathophysiology of severe lung oedema

The experimental evidence is that the lung can withstand for a long time in a condition of interstitial oedema but the transition to severe oedema is a tumultuous event developing within minutes [9]. We identified the sequence of events leading to the increase in extravascular lung water [9]. A sustained condition of interstitial oedema indeed results in loss of integrity of the native architecture of the large proteoglycans of the interstitial matrix and also of the intermediate sized proteoglycans of the basement membrane [9]. Therefore, as a result of fragmentation, two important features are lost, namely the high tissue elastance and the low permeability of the capillary endothelium. A combination of these two effects leads to the “accelerated phase” of development of severe lung oedema. More specifically, the increase in tissue compliance leads to a decrease in Pi (fig. 7⇑), thus, restoring a transendothelial Starling pressure gradient. However, although this gradient is smaller than that occurring in physiological conditions, it causes an unopposed increase in microvascular filtration due to the increase in microvascular permeability. In the so called “hydraulic oedema” induced by venous fluid overloading, the degradation process initially involves large matrix proteoglycans, while in the lesional type of oedema (as in acute pancreatitits) there is a major initial degradation of proteoglycans of the basement membrane. Thus, there may be a variable contribution to oedema formation due to the loss of “tissue safety factor” and/or to the increase in microvascular permeability. Despite these differences, we provide a common pathophysiological interpretation for the development of severe oedema as this occurs when a critical threshold is overcome in the process of proteoglycan fragmentation [9]. Factors contributing to progressive disorganisation/fragmentation of the proteoglycan mesh include: 1) mechanical yielding; 2) weakening of the noncovalent bonds of proteoglycans with other matrix components at link sites due to increased hydration/distance; and 3) the activation of tissue matrix metalloproteases (MMP)2 and MMP9, as is specifically the case for hypoxia exposure [34].

The increase in permeability leading to alveolar flooding may occur through porosity of the paracellular pathway of the order of 50–100 nm, which easily allows passage of albumin. Red blood cells can also be found in the alveolar fluid and this reflects major lesion of the air–blood barrier.

It is of interest to discuss the model of control of extravascular lung fluid considering the case of the transplanted lung which, of course, is deprived of lymphatics, whose regeneration might require ∼3 weeks [35]. Pulmonary oedema is reported in ∼55% of the cases [36]; however, in a good percentage of cases, the transplanted lung can act as a good diffuser essentially revealing that the state of the extracellular matrix can assure a tissue safety factor. The major complication within the first 24–48 h following transplantation is severe oedema (∼10% of the cases) attributed to ischemia–reperfusion injury [37]. A good body of literature focuses on the central role of neutrophil activation [37], which induces reactive oxygen species causing a major increase in microvascular permeability, diffuse alveolar damage and inhibition of the active alveolar fluid reabsorption [38]. Fragmentation of extracellular matrix and lack of clearance of the fragments is involved in neutrophil activation [39]. Lung preservation and reperfusion techniques were in fact revised in order to prevent ischemia–reperfusion injury [40].

Several factors may interact to influence the severity of lung oedema (all in equation 1). The most potent factor is the increase in Kf (capillary surface area × permeability, equation 1). All conditions leading to an increase in cardiac output cause an increase in capillary surface area due to capillary recruitment, such as hypoxia in strenuous exercise. Left heart failure also causes an increase in capillary surface area. Hypoxia [34, 41], inflammatory states [42] and increase in shear rate at capillary level [43] also cause an increase in permeability. Shear rate increases when flow velocity increases with an increase in cardiac output (such as in hypoxia/exercise) or a decrease in vascular bed (such as after lobectomy). The other factor involved in oedema formation is the increase in capillary pressure above the physiological value of ∼10 cmH₂O. This is considered to occur as a complication of pulmonary hypertension, the hypothesis being that sudden release of the arteriolar vasomotor tone might locally induce an increase in capillary pressure.

Recovery from lung oedema

Unlike pleural effusion, the recovery from lung oedema may be a question of hours. This requires redeposition and remodelling of the extracellular matrix, in particular as far as the proteoglycan family is concerned. A key, still unsolved, issue is the understanding of how the cell-tissue interaction influences either the reparative process or the development of a severe form of lung oedema such as acute respiratory distress syndrome. Thus, it might be useful to identify interstitial oedema as a sharp edge between tissue repair or manifestation of a severe disease. There is evidence that removal of matrix fragments is critical for successful repair [44]; in particular failure to clear hyaluronan fragments leads to “unremitting” inflammation [45].

We explored the cellular-tissue mechanical interaction evoked by the increase in parenchymal stresses. Interstitial lung oedema was found to cause a considerable increase of mRNA transcripts for tumour necrosis factor-α, interleukin-6 and fibroblast growth factor-β [46]. Tissue remodelling has already been described in the phase of interstitial lung oedema and is considered to be a result of the signalling–transduction processes initiated by endothelial and epithelial cells. In fact, signs of activation of these cells corresponding to the phase of increased Pi were demonstrated by differential expression of lipid microdomains, either caveolae or lipid rafts, which represent specialised signalling–transduction platforms on the plasma membrane. The expression of caveolae increased substantially in response to cardiogenic lung oedema [47–49] which causes major initial damage to large matrix proteoglycans. However, the expression of lipid rafts was increased in response to hypoxic oedema [50], which induces major initial fragmentation of heparansulphate proteoglycans of the basement membrane. Remodelling of the expression of lipid microdomain seems to be a rather fast response based on relocation of available constituents on the plasma membrane surface [51]. It appears tempting to think of these differential activations as fine mechanisms tuned to promote a specific reparative process to resist oedema formation. Endothelial and epithelial cells may indeed act as good sensors of changes in force/pressure applied on their surface to elicit a signalling-mechano-transduction response [49]. In fact, they are highly deformed in situ due to attachments to adjacent cells and to the basement membrane. Thus, because of these physical constraints they act as rigid structures, an important feature for mechano-transduction [49].

Early diagnosis of lung oedema

As mentioned previously, interstitial oedema represents a potential unstable condition between tissue repair and severe tissue lesion and, in clinical practice, it is still critical to detect markers of impending risk of developing severe oedema. Based on the notion that interstitial oedema increases the parenchymal stresses, we monitored the changes in lung mechanics induced by an experimental model of lung oedema by forced oscillation technique. We found that reactance measured at a low frequency (4–5 Hz) decreased progressively and significantly over time when interstitial oedema was present and water accumulation in the interstitial compartment was steadily maintained within 10% of control value [52]. No change in small airways patency was demonstrated so that the finding could be attributed to changes in mechanical properties of the lung tissue under conditions of increased parenchymal stresses. It was concluded that monitoring of reactance could have a potential impact in clinical practice as an early marker of developing frank oedema, before any change in lung compliance or alveolar fluid accumulation occurs.