For generations, the axiom guiding transcapillary fluid behavior was Starling’s model, first proposed in 1896 [26]. The model expresses fluid flux as a balance between opposing hydrostatic and oncotic pressures. Along the length of the capillary filtration is favored at the arteriolar end and reabsorption at the venular end.

However, in vitro and in vivo deviations from the classic Starling principle have been noted [27], such as absence of the venous reabsorption [28] and lymphatic flow [29] required to prevent interstitial edema, and lack of importance of the interstitial colloid osmotic pressure in determining transendothelial fluid balance [30]. This led to further investigation into non-Starling mechanisms of barrier regulation involving the endothelial glycocalyx layer (EGL) [31].

Danielli first proposed the existence of the EGL in 1942 [32]. It is a dynamic, fragile, and complex layer of membrane-bound macromolecules at the luminal surface of the vascular endothelium [31]. The composition and thickness of the glycocalyx change constantly, as it is continually sheared by plasma flow and replaced [33]. Its components have a net negative charge and therefore repel negatively charged molecules and blood cells [34].

A primary function of EGL is to regulate and influence vascular permeability [35]. Together with circulating substances, it forms a barrier that prevents circulating cells and macromolecules from entering the interstitium. In contrast to the original Starling model, which explained regulation of fluid balance occurring across the entire endothelial cell, a revised model has been proposed whereby the hydrostatic and osmotic forces act only across the EGL surface layer on the luminal aspect of the endothelium [30]. These forces reach equilibrium very quickly, resulting in a much lower fluid flux than predicted by the traditional Starling equation (see Fig. 5.1).

The EGL has other functions. It regulates blood cell-endothelial interaction by its negative charge and via specific adhesion molecules for leukocytes and platelets. These are normally hidden deep within the glycocalyx structure, but become exposed following damage to the EGL [35]. It also protects the vascular endothelium from shear stress and oxidative damage, via nitric oxide-induced vasodilation [36] and scavenging of oxygen free radicals [34].

The EGL may be injured by inflammatory cytokines [37], surgical trauma, and ischemia-reperfusion [38]. Hypervolemia damages the EGL, both by dilution of plasma proteins and via release of atrial natriuretic peptide, which strips the EGL [38]. Loss of the intact EGL causes increased vascular permeability and fluid extravasation. Loss of plasma proteins further compounds this. Leukocyte adhesion molecules are exposed, promoting cellular adhesion, migration, and further inflammation [39]. This vicious cycle of increased permeability, extravasation, and inflammation leads to pulmonary edema, as is observed in ALI.

Several empiric strategies, based on animal experiments, have been proposed to protect the EGL, including avoiding hypervolemia, albumin infusion [40], corticosteroids [41], antithrombin III [42], and direct inhibitors of inflammatory cytokines [43]. Volatile anesthetic agents, when compared with propofol, have been associated with less local release of inflammatory mediators [44, 45] and less glycocalyx destruction [46] (see Fig. 5.2).

The alveolar endothelium also plays a role in the regulation of pulmonary interstitial fluid balance. Fluid transport across the endothelium may occur via tight junctions, breaks in tight junctions and vesicular transport [47]. Leaky junctions are associated with cell death and allow passage of larger molecules. Epithelial sodium channels (ENaCs) are able to enhance the clearance of alveolar fluid [47], and they may be stimulated by beta-adrenergic agonists [48].

Endothelial damage has therefore been implicated in the pathogenesis ALI after lung resection surgery. Endothelial injury maybe induced by activation of systemic and local inflammatory mediators, related to positive pressure ventilation; “volutrauma;” oxygen toxicity; ischemia-reperfusion injury; surgical trauma; and preexisting lung disease [1, 3]. Endothelial cell injury results in disruption of intercellular endothelial cell junctions, cytoskeleton contraction, and cell death, leading to increased permeability of the alveolar-capillary barrier and decreased lung compliance [1].

Although fluid overload is well recognized as a risk factor for ALI after thoracic surgery, there also appear to be other factors at play. ALI may still occur when very strict fluid-restrictive strategies are implemented [49].

Lung lymphatics play a key role in fluid clearance from the lung [47]. Capillary filtrate that enters the interstitium is drained by lymphatics, and when their capacity to drain fluid is exceeded, pulmonary edema will occur [3]. Although lymph flow can increase sevenfold in response to elevated interstitial pressure [50], in the perioperative setting, this capacity may be reduced. Surgical trauma related to lung resection surgery is thought to be an important factor influencing this [3]. Pulmonary lymphatic drainage is not symmetrical: the drainage of the right lung is essentially ipsilateral (>90 %), whereas the left lung has a significant contralateral contribution (>55 %) [51]. Therefore, right pneumonectomy confers a significantly higher risk of pulmonary edema in the left lung, as over half its lymphatic drainage will be lost, which has been demonstrated clinically [7, 17].

Lymphatic drainage may be further impaired by postoperative right ventricular (RV) dysfunction: the resultant elevation in central venous pressure will reduce the drainage capacity of the lymphatic system [52]. RV dysfunction is very common after lung resection surgery, particularly pneumonectomy [53], and is thought to relate to increased RV afterload and tachycardia [54, 55].

Many goal-directed protocols target cardiac index, which may be measured using a variety of modalities, including the pulmonary artery catheter, transpulmonary thermodilution (e.g., PiCCO monitor), pulse contour analysis (e.g., FloTrac-Vigileo system), and transesophageal Doppler measurement. Although use of esophageal Doppler is impractical in esophageal surgery, it has been used with good effect in lung resection surgery. Esophageal Doppler was able to detect a reduction in stroke volume index in lung resection surgery, despite unchanged heart rate and mean arterial pressure, and was used to guide hemodynamic support and fluid therapy [81].

SVV and PPV both use the heart-lung interaction, integrating the effects of preload, respiratory variation, and blood pressure to assess fluid responsiveness [3, 47]. There are some theoretical limitations to the use of these indices during thoracic surgery. Firstly, there has been concern regarding their validity during open chest conditions [82]. Secondly, due to the dependence of these measurements on respiratory variation, their accuracy also is dependent on tidal volume. At relatively large tidal volumes of 8–10 mL/kg during two lung ventilation (TLV), SVV ≥ 12 ml/kg and PPV ≥ 10 mL/kg correlate highly with fluid responsiveness [83]; however, ventilation with a lung protective strategy may not have the same correlation. Thirdly, the volume of shunted blood through the non-ventilated lung should not contribute to the generation of SVV and PVV, necessitating a lower threshold value during OLV than that used during TLV [3, 84]. In one study examining SVV during OLV, it was shown only to be acceptably predictive of volume responsiveness with tidal volumes >8 mL/kg, with a threshold for fluid responsiveness of 10.5 % [85]. At the upper limits of fluid responsiveness, small increases in cardiac output are associated with large increases in extravascular lung water. This is particularly a problem in conditions of increased endothelial capillary permeability (see Fig. 5.3) [84].