Bench-to-bedside review: Ventilation-induced renal injury through systemic mediator release - just theory or a causal relationship?

We review the current literature on the molecular mechanisms involved in the pathogenesis of acute kidney injury induced by plasma mediators released by mechanical ventilation. A comprehensive literature search in the PubMed database was performed and articles were identified that showed increased plasma levels of mediators where the increase was solely attributable to mechanical ventilation. A subsequent search revealed articles delineating the potential effects of each mediator on the kidney or kidney cells. Limited research has focused specifically on the relationship between mechanical ventilation and acute kidney injury. Only a limited number of plasma mediators has been implicated in mechanical ventilation-associated acute kidney injury. The number of mediators released during mechanical ventilation is far greater and includes pro- and anti-inflammatory mediators, but also mediators involved in coagulation, fibrinolysis, cell adhesion, apoptosis and cell growth. The potential effects of these mediators is pleiotropic and include effects on inflammation, cell recruitment, adhesion and infiltration, apoptosis and necrosis, vasoactivity, cell proliferation, coagulation and fibrinolysis, transporter regulation, lipid metabolism and cell signaling. Most research has focused on inflammatory and chemotactic mediators. There is a great disparity of knowledge of potential effects on the kidney between different mediators. From a theoretical point of view, the systemic release of several mediators induced by mechanical ventilation may play an important role in the pathophysiology of acute kidney injury. However, evidence supporting a causal relationship is lacking for the studied mediators.


Introduction
Acute kidney injury (AKI) is a common problem in critically ill patients and carries signifi cant morbidity and mortality. Based on a recent multinational study, the incidence of AKI is estimated to be 5.7%, with a mortality of 60% [1]. AKI rarely occurs in isolation but usually develops in the context of multiple organ failure. Despite advances in dialysis technology and supportive care, mortality resulting from AKI has remained unchanged over the past years and is as high as 80% when associated with respiratory insuffi ciency [1,2]. An observational study recently found that 75% of all patients with acute respiratory failure required some form of renal replacement therapy [1].
Mechanical ventilation (MV) is an independent risk factor for the development of AKI and can contribute to its development by three proposed mechanisms: blood gas disturbances leading to hypoxemia or hypercapnia and subsequent neurohumoral-mediated eff ects on renal blood fl ow during MV; changes in cardiac output, redistribution of intra-renal blood fl ow and stimulation of hormonal and sympathetic pathways may aff ect systemic and renal hemodynamics, thereby decreasing renal blood fl ow; and MV-induced biotrauma, defi ned as a pulmonary infl ammatory reaction to MV with pulmo nary mediator release [1,3]. Subsequent spill-over of these mediators into the systemic circulation may contribute to AKI [4].
Although various processes play signifi cant roles in the pathophysiology of AKI, this review focuses specifi cally on the potential role of plasma mediators released as a result of MV in the pathogenesis of AKI. First, we review the current clinical and experimental literature describing mediators that are systemically released during MV and their eff ect on the kidney. Th e causality of the relationship between systemically released mediators and AKI will be explored. Second, we identify mediators whose release is attributable to MV and discuss the potential eff ects of these mediators on the kidney. Th is will provide a framework for future research on ventilation-induced renal injury through systemic mediator release.

Abstract
We review the current literature on the molecular mechanisms involved in the pathogenesis of acute kidney injury induced by plasma mediators released by mechanical ventilation. A comprehensive literature search in the PubMed database was performed and articles were identifi ed that showed increased plasma levels of mediators where the increase was solely attributable to mechanical ventilation. A subsequent search revealed articles delineating the potential eff ects of each mediator on the kidney or kidney cells. Limited research has focused specifi cally on the relationship between mechanical ventilation and acute kidney injury. Only a limited number of plasma mediators has been implicated in mechanical ventilation-associated acute kidney injury. The number of mediators released during mechanical ventilation is far greater and includes pro-and anti-infl ammatory mediators, but also mediators involved in coagulation, fi brinolysis, cell adhesion, apoptosis and cell growth. The potential eff ects of these mediators is pleiotropic and include eff ects on infl ammation, cell recruitment, adhesion and infi ltration, apoptosis and necrosis, vasoactivity, cell proliferation, coagulation and fi brinolysis, transporter regulation, lipid metabolism and cell signaling. Most research has focused on infl ammatory and chemotactic mediators. There is a great disparity of knowledge of potential eff ects on the kidney between diff erent mediators. From a theoretical point of view, the systemic release of several mediators induced by mechanical ventilation may play an important role in the pathophysiology of acute kidney injury. However, evidence supporting a causal relationship is lacking for the studied mediators.

Methods
We performed an extensive literature search in PubMed using medical subject headings and text words, supplemen ted by scanning the bibliographies of the recovered articles. We combined 'acute renal failure' and 'acute kidney injury' using the term 'OR' . Th is search was subsequently combined with 'mechanical ventilation' using the Boolean operator ' AND' . Using a similar search strategy, using 'mediator' and 'cytokine' we identifi ed 19 diff erent plasma mediators that increased during MV. We included only in vivo studies in which the increase in plasma mediator levels was exclusively attributable to MV. We excluded neurohumorally increased mediators, mediators increased in renal tissue samples and mediators derived from in vitro experiments exposing cell cultures to mechanical stretch. Each mediator was searched in PubMed, also including alternative names and abbreviations. We combined these results with the terms 'glomerular' , 'glomerulus' , 'tubular' , 'mesangial' , 'mesangium' , 'podocyte' , 'acute renal failure' and 'acute kidney injury' . To delineate the potential eff ects of these mediators on the kidney, we limited the articles to studies that solely studied eff ects on the kidney or on diff erent kidney cell types.

Mechanical ventilation, systemic mediator release and the kidney
Th e importance of MV in morbidity and mortality of patients suff ering from acute respiratory distress syndrome is stressed by the 2000 landmark study by the ARDS Network. In this multi-center trial, lung protective ventilation decreased morbidity and mortality rates compared to a conventional strategy [5]. Although the exact mechanisms remain unknown, the biological response of the lungs to the eff ects of MV was aptly named biotrauma to describe the ongoing changes in pulmonary infl ammation and the systemic release of infl ammatory media tors [6]. Th e biotrauma hypothesis is supported by evidence from experimental models ranging from mechanically stressed cell systems to isolated lungs, intact animals, and humans [7]. Various mechanisms are responsible for the ventilation-induced release of media tors. Th ere are four principal mechanisms, all of which appear to be clinically relevant: stress failure of the plasma membrane (necrosis); stress failure of endothelial and epithelial barriers (decompartmentalization); over disten sion without tissue destruction (mechano trans duction); and eff ects on vasculature, independent of stretch and rupture [8]. Th e possible eff ects of systemically released and circulating mediators during ventilator-induced lung injury on organs distant from the lungs has prompted research to focus on the potential eff ects of mediators on the kidney (Table 1). Th us far only one clinical study has compared a conventional MV strategy with a lung-protective strategy in patients with acute respiratory distress syndrome. Th is single-center study found increased kidney failure in the conventional strategy group. A correlation was found between plasma IL-6 levels and the number of failing organs in the same patients [9]. In these patients an association between plasma soluble Fas ligand (sFasL) levels and changes in creatinine was also found [10]. Th e authors conclude that mediator release during MV is correlated to the development of multi-organ failure and these fi ndings may partially explain the decrease in morbidity and mortality in patients ventilated with a lung protective strategy. In animal experiments the role of MV on the kidney was further explored, focusing on the role of pro-infl ammatory mediators [11][12][13][14], renal apoptosis [10], vasoactive mediators and renal blood fl ow [15], coagulation and fi brinolysis [16,17], and other mediators such as vascular endothelial growth factor (VEGF) ( Table 1) [18]. Of specifi c importance is the previously mentioned study by Imai and colleagues [10]. In contrast to other animal studies that are observational by nature, this is the only study that used specifi c blocking of mediators, albeit in vitro. Imai and colleagues found increased renal apoptosis after injurious ventila tion in rabbits. In vitro blocking of sFasL prevented induction of apoptosis of cultured kidney cells by plasma from rabbits ventilated with an injurious ventilatory strategy [10]. Although limited in number, studies linking MV with acute respiratory failure have discovered several new potential pathways in addition to pro-infl ammatory pathways through which MV may cause acute respiratory failure. To date, however, no study has established a causal relationship between specifi c mediators and acute respiratory failure during MV in vivo through, for instance, intervention studies where the release of mediators is prevented or by blocking released mediators. Table 2 shows the mediators whose increased release was attributable to MV. Several clinical studies identifi ed a large number of plasma mediators [5,[19][20][21][22][23][24]. Th ese plasma mediators are not only pro-infl ammatory by nature, but anti-infl ammatory mediators have been identi fi ed as well [22][23][24]. Furthermore, identifi ed media tors are also involved in coagulation, fi brinolysis, cell adhesion and surfactant homeostasis [5,[19][20][21][22][23][24][25][26]. Most research has focused on pro-and anti-infl ammatory mediators as well as chemokines, and only limited studies have outlined a role for mediators primarily involved in cell growth and diff erentiation or apoptosis (Table 2) [10,18]. A more indepth analysis of the various media tors summarized in Table 2 may provide new therapeutic insights.
Both in vitro and in vivo studies have shown that TNFα induces caspase 8-dependent apoptosis of renal tubular cells and renal endothelial cells by binding to either the TNF-receptor-1 or Fas-receptor [50][51][52][53]. In glomerular endothelial cells, cytochrome c infl ux in the cytosol was found after TNF-α stimulation, suggesting a role in the mitochondrial pathway as well [54]. Ceramide is an important signaling molecule in cellular responses, includ ing apoptosis [55]. In both mesangial cells and glomerular endothelial cells TNF-α stimulated ceramide formation, but only in the latter did this lead to increased apoptosis [56,57]. TNF-α also suppresses expression of anti-apoptotic proteins, both in vitro and in vivo [53,54]. Hruby and colleagues [58] showed in vitro that TNF-α induced cytolysis in mesangial cells, but not in glomerular epithelial cells. Th is may be partially attributable to the TNF-α-induced production of reactive oxygen species by mesangial cells [59]. TNF-α stimulates mesangial cells to produce a variety of vasoactive mediators in vitro (Table 3) [60][61][62][63][64][65][66][67][68][69]. Piepot and colleagues [70] showed impaired endotheliumdependent arterial relaxation after TNF-α exposure. By disturbing the balance between vasoconstriction and vasodilatation, TNF-α is thus potentially capable of reducing glomerular blood fl ow and glomerular fi ltration rate [71]. Additionally, TNF-α induced downregulation of angiotensin (Ang)-II type-1 receptor expression in vivo, which may explain the frequently observed vasodilation during sepsis [72].
TNF-α-induced histologic kidney damage is frequently characterized by glomerular fi brin deposition. By stimulat ing renal plasminogen activator inhibitor (PAI)-1 gene expression and increasing the production of tissue factor by mesangial and endothelial cells, TNF-α can contribute to fi brin deposition [75,76].
In a series of animal experiments, Schmidt and colleagues [77][78][79][80] hypothesized a role for urea, glucose, sodium and chloride transporters in sepsis-associated tubular dysfunction. Within hours after lipopolysaccharide injection in mice they showed decreased renal blood fl ow and glomerular fi ltration and impaired tubular sodium handling associated with decreased levels of the aforementioned transporters. In TNF-α-challenged mice they found signifi cant downregulation of genes coding for urea, glucose, sodium and chloride transporters as well as for chloride channels and Na + /K + -ATPase-α 1 compared to wild-type mice.
In proximal tubule cells in vitro, TNF-α caused a downregulation of gene expression of the nuclear hormone receptor liver X receptor/retinoid X receptor (LXR) and several of its target genes and coactivators [81]. During the acute phase TNF-α interferes with lipid metabolism in the kidney, with potential subsequent eff ects on the anti-infective and anti-infl ammatory properties of lipids [82].
IL-1β is also involved in hemodynamic instability during septic shock. By downregulating Ang-II type-I receptors, IL-1β may be partially responsible for the decreased reactivity to vasoconstrictors [72]. Additionally, in mesangial cells, IL-1β, through inducible nitric oxide synthase (iNOS), caused increased production of NO, known for its vasodilatatory eff ects, and prostaglandin (PG)E 2 , which has potential vasodilatory properties [39,94,95].
Data on the eff ects of IL-1β cell proliferation are limited. Tateyama and colleagues [96] showed that IL-1β could function as an autocrine growth factor for rat glomerular epithelial cells in vitro.
IL-1β can aff ect both coagulation and fi brinolysis in vitro. In mesangial cells, IL-1β upregulated tissue factor expression by a protein kinase C-dependent pathway, with an eff ect on tissue factor activity only when cells were rendered apoptotic [97]. Also in these cells, IL-1β was capable of inducing the fi brinolytic enzyme tissue type plasminogen activator (tPA), but also its inhibitor PAI-1 [98]. Th e eff ects of IL-1β on coagulation and fi brinolysis in the kidney in vivo remain unknown.
In the aforementioned series of animal experiments, Schmidt and colleagues [77][78][79][80] also found that IL-1βchallenged mice signifi cantly downregulated genes coding for urea, glucose, sodium and chloride trans porters as well as for chloride channels and Na + /K + -ATPase-α 1 compared to wild-type mice. Th is implicates IL-1β in sepsis-associated tubular dysfunction with decreased glomerular fi ltration rate, failure of urine concentration, decreased urine osmolality, increased fractional sodium excretion and glucosuria.
In human proximal tubule cells in vitro, IL-1β, like TNF-α, caused downregulation of gene expression of the nuclear hormone receptor LXR and several of its target genes and also of its coactivators [81]. Th e authors suggest a role for IL-1β in lipid metabolism in the kidney during the acute phase.

Interleukin-6
Th e exact nature of IL-6 remains the subject of debate -it has been extensively described as both pro-and antiinfl am matory [99,100]. IL-6 levels increase during hypoxia, tissue damage and organ failure [101][102][103] and predict mortality in patients with acute renal failure [104]. In the kidney, IL-6 is involved in infl ammation, leukocyte adhesion and infi ltration, apoptosis and survival, vasoactivity, prevention of oxidative stress, cell proliferation and lipid homeostasis during the acute phase (Table 3).
Th e role of IL-6 in apoptosis and survival is complex. In cisplatin-induced renal failure, mice lacking IL-6 had better survival rates despite decreased renal function. Th is was associated with upregulation of both pro-and anti-apoptotic genes [108]. Th e authors explain these phenomena by the fact that the upregulation of proapoptotic genes disappears after 24 hours, while antiapoptotic genes remain upregulated for 72 hours [108]. Others also found a positive eff ect on survival in IL-6 knock-out mice with improved renal function [105,106]. In mice exposed to intravenous IL-6, Schmidt and colleagues [72] found decreased expression of Ang-II type-I receptors, which are involved in vasoconstriction, potentially explaining vasodilatation during shock. In an I/R model and a mercury chloride-induced model of acute renal failure, IL-6 knock-out mice and mice treated with anti-IL-6 antibodies had lower levels of oxidative stress and NO-dependent oxidative stress. In addition, IL-6 bound to soluble IL-6 receptor, likely shedded from neutrophils during AKI, increased gene expression of heme oxygenase-1 and restriction factor-1, both known to protect against oxidative stress [105,106].
Similar to TNF-α and I L-1β, Schmidt and colleagues [77][78][79] hypothesized a role for urea, glucose and chloride transporters in sepsis-associated tubular dysfunction with failure of urine concentration, decreased urine osmo lality and glucosuria. In IL-6-challenged mice they found signifi cant downregulation of genes coding for urea, glucose and chloride transporters as well as for chloride channels and Na + /K + -ATPase-α 1 expression compared to wild-type mice, though to a lesser extent than TNF-α and IL-1β.
Poloxamer 407 induces hyperlipidemia that protects against renal I/R dysfunction. Th is was associated with decreased plasma levels of IL-6, but recombinant IL-6 infusion abrogated these eff ects. Th ese results were confi rmed in apolipoprotein-E-and angiopoietin-like 3-defi cient mice, which suff er from hypercholesterolemia and hypolipidemia, respectively [115].

Interleukin-10
IL-10 exerts its anti-infl ammatory eff ect through inhibition of MHC class II-associated antigen presen tation and by decreasing circulating levels of CC and CXC chemokines [116]. However, in a dose-dependent manner, IL-10 can also promote infl ammation through eff ects on B cells and natural killer cells and by stimulating cytokine production [116]. Increased levels of IL-10 predicted mortality in patients with acute renal failure and, interest ingly, patients with specifi c IL-10 gene polymorphisms required less renal support during sepsis from pneumonia [104,117]. Th e eff ects of IL-10 on the kidney involve eff ects on infl ammation, infl ammatory cell recruitment and infi ltration, apoptosis, necrosis and cell cycle activity, vasoactivity and cell proliferation (Table 3).
In vivo studies in diff erent models of AKI, intra venous IL-10 administration decreased TNF-α produc tion and prevented creatinine increase in mice [118]. Mesangial cells in vitro showed less production of TNF-α and IL-1β after stimulation with lipopolysaccharide in the presence of IL-10 [119]. Deng and colleagues [118] reported decreased ICAM-1 expression in mice during kidney injury and IL-10 injection; histological analysis also revealed decreased cast formation and leukocyte infi ltration. Contradictory to these fi ndings was an observation made by Chadban and colleagues [120] showing increased ICAM-1 expression on rat mesangial cells after stimulation with IL-10. In mice, administration of IL-10 prevented both apoptosis and necrosis, mainly in the outer stripe of the kidney, after cisplatin and I/R-induced kidney injury. IL-10 also decreased cell cycle activity [118].
IL-10 addition to glomerular epithelial cells in vitro reduced VEGF, a potent modulator of capillary permeability [121]. In vivo, IL-10 decreased iNOS, and in in vitro culture of cortical tubule cells, IL-10 exposure decreased nitrite formation [118].
In vitro experiments showed proliferation-stimulating eff ects of IL-10 on mesangial cells through platelet derived growth factor receptor α and β, but also up regulation of IL-10 mRNA, suggesting a possible autocrine mechanism [122][123][124]. Th ese in vitro results were confi rmed in vivo [124].

Interleukin-1 receptor antagonist
IL-1 receptor antagonist (IL-1RA) is a physiological inhibitor of IL-1β activity through competitive binding to the IL-1β receptor. Recombinant IL-1RA administration during sepsis showed a mortality benefi t of almost 5% [131]. In mesangial cells addition of IL-1RA decreased gelatinase B, stromelysin, MCP-1 and IL-8 RNA and protein levels after stimulation with IL-1α and IL-1β [132,133]. In in vivo models of anti-glomerular basement membrane antibody glomerulonephritis and renal I/R treatment with IL-1RA resulted in improved kidney func tion, decreased expression of ICAM-1 and reduced renal histological damage, including decreased infi ltration of lymphocytes, neutrophils and macrophages and less apoptosis (Table 3) [134][135][136][137][138].

Interleukin-8
Th e chemokine IL-8 is produced mainly by macrophages, but also by renal tubular epithelial cells, mesangial cells and podocytes [139][140][141][142]. IL-8 levels predict the development of AKI, duration of MV and mortality in patients with AKI [104,143,144]. Exposure of mesangial cells to IL-8 in vitro leads to selective expression of cyclooxygenase (COX)1, but not COX2, and subsequent synthesis of PGE 2 [145]. In vivo infusion of IL-8 in rats causes increased albuminuria, mediated through alterations of sulfate metabolism by the glomerular basement membrane (Table 3) [146].

Macrophage infl ammatory protein-2
Macrophage infl ammatory protein (MIP)-2, a member of the superfamily of chemokines, is a potent chemotactic factor for neutrophils and stimulates the production of other infl ammatory mediators such as IL-1β and TNF-α [147]. In kidneys, mesangial cells and glomerular epithelial cells stimulated by NO or IL-1β are capable of synthesizing MIP-2 [148][149][150]. Exposure of mesangial cells to MIP-2 in vitro stimulates the release of MCP-1, RANTES and also MIP-2 [151]. Specifi c blocking of MIP-2 in in vivo models of shiga toxin-induced renal infl ammation and anti-glomerular basement membrane antibody glomerulonephritis prevented renal neutrophil infl ux and fi brin deposition and decreased proteinuria (Table 3) [152,153].

Keratinocyte chemoattractant
Keratinocyte chemoattractant (KC), also known as growth related oncogene or CXCL1, in mesangial cells increased production of pro-infl ammatory mediators such as MCP-1, RANTES, MIP-2 and KC [151]. KC exhibited neutrophil-attracting properties as shown in a mouse model of shiga toxin-induced renal injury [153]. In vitro KC stimulated proliferation of inner medullary collecting duct cells [154]. KC also stimulates mesangial cells to produce COX1 and enhances PGE 2 synthesis (Table 3) [145].

Coagulation and fi brinolysis Plasminogen activator inhibitor-1
In patients with acute lung injury, PAI-1 is a prognostic factor for the development of acute renal failure [126]. In PAI-1 knock-out models or models overexpressing PAI-1 with anti-glomerular basement membrane glomerulonephritis, PAI-1 increases leukocyte infi ltration, crescent formation, fi brin deposits, fi bronectin synthesis and collagen accumulation [169][170][171]. Similar results were found in rodents with unilateral ureteral obstruction with increased transforming growth factor-β1 levels and decreased levels of urokinase [172]. Functionally, PAI-1 knock-out mice showed decreased albuminuria in a diabetes model (Table 3) [173].

Activated protein C
In patients with severe sepsis, baseline activated protein C (aPC) levels were inversely associated with worsening renal function and/or subsequent dialysis and treatment, whereas treatment with aPC was associated with improved renal function [179]. Th e anti-infl ammatory eff ects of aPC are shown in the downregulation of the expression of TNF-α, IL-6, IL-8 and IL-18 [180][181][182]. By decreasing KC and MIP-2 protein and MCP-1 mRNA, aPC potentially prevents infl ammatory cell recruitment [181][182][183]. Additionally, aPC suppresses leukocyte rolling and adhesion [184]. Histologically this leads to decreased leukocyte infl ux and also decreased renal myelo peroxidase levels [180,181]. In the same histological specimens, aPC prevented renal necrosis and apoptosis of glomerular and endothelial cells and podocytes [180][181][182][183][184][185]. In a model of diabetes, Isermann and colleagues [185] found aPC to have an antioxidant eff ect, decreasing nitrosative stress by decreasing kidney nitrotyrosine levels. Positive hemodynamic eff ects have been observed in various models, whereby aPC increased renal blood fl ow and peritubular fl ow, potentially by the observed decrease in adrenomedullin, iNOS, angiotensinogen mRNA, Ang converting enzyme and Ang-II [184]. Vascular permeability is also decreased by aPC [180,181]. Th e anticoagulant properties of aPC are highlighted by a decrease in circulating fi brin degradation products and decreased extracellular matrix depositions [180,185]. Combined, these eff ects of aPC preserve renal function as measured by creatinine, blood urea nitrogen levels and proteinuria (Table 3) [181][182][183][184].

Vascular endothelial growth factor
In a rodent model of ventilator-induced lung injury, Choi and colleagues [18] indicated a role for VEGF in endothelial NOS-mediated vasopermeability in lungs and kidneys. VEGF is a potent endothelial cell mitogen, promotes endothelial cell diff erentiation and survival, stimulates angiogenesis and enhances vascular permeability. While deleterious in some forms of renal disease, VEGF may contribute to recovery in others [186,187].
In rodent models of glomeruolonephritis, intravenous VEGF decreased MCP-1 and ICAM-1 levels with a subsequent decrease in infi ltrating leukocytes [188]. VEGF prevented glomerular and tubulointerstitial cell apoptosis and necrosis in models of hemolytic uremic syndrome and mesangio-proliferative nephritis, which was confi rmed in vitro [189][190][191][192]. In vivo VEGF stimulated endothelial nitric oxide synthase expression in rats in a remnant kidney model, and in glomerular endothelial cells this increased NO expression [193,194]. One of the characteristics of VEGF is enhancement of vascular permeability; this was confi rmed in vitro [195,196]. Th e proliferative eff ects of VEGF have been well described, both in vivo and in vitro, but these properties are likely of limited interest for the development of AKI (Table 3).

Soluble Fas ligand
sFasL is up to 1,000-fold less active than membranebound FasL in inducing apoptosis and has even been suggested to have antagonistic properties [197,198]. However, Imai and colleagues [10] showed apoptotic activity of serum of mechanically ventilated rabbits on renal tubular cells in vitro, which could be blocked by an anti-sFasL Fas:Ig fusion protein (Table 3).

Mediators -theory or causal relationship?
Central in the biotrauma hypothesis is the increase in intra-pulmonary mediator levels and the spill-over of these mediators from the lung into the systemic circu lation. Several mediators are systemically increased during MV (Tables 1 and 2), although the exact cellular origins of the systemically measured mediators remains unknown [7]. Most of the mediators increased during MV have potential and well described eff ects on the kidney.
Most studies have focused on pro-infl ammatory and chemotactic mediators, especially TNF-α, IL-6 and MIP-2. Th e eff ects of these pro-infl ammatory mediators on the kidney have been studied to varying degrees. In vivo evidence indicates that TNF-α can cause and contributes to AKI, in contrast to IL-1β and MIP-2, for which suffi cient in vivo data are lacking. IL-6 has been shown to be involved in AKI, but confl icting reports exist and no defi nite conclusion can be drawn. Much is known about the eff ects of anti-infl ammatory and chemotactic media tors on the kidney. Strong evidence indicates a protective role for anti-infl ammatory mediators, especially IL-10, in the development of AKI, but insuffi cient evidence exists to indicate a direct role in AKI for chemotactic mediators. Little is known about the potential eff ects on the kidney of some frequently studied mediators released during MV, such as soluble ICAM, soluble VCAM and sFasL. Th e potential of these mediators will remain unknown in the absence of studies on their possible eff ects on the kidney. Other mediators, for example, aPC and VEGF, have several well known eff ects on the kidney but have received little attention in studies on mediator release during MV. Studies of aPC and VEGF during MV have great potential to further delineate the eff ects of these mediators on the kidney. IL-10, IL-8 and aPAI-1 all have predictive value for the development of AKI or AKI-associated mortality. Despite this, few studies have focused on the role of MV and, even though their increase may be an epiphenomenon, little is known about their potential role in the pathophysiology of AKI. Interestingly, we did not identify studies using specifi c blocking of mediators in vivo or specifi c knock-out models that could establish a causal relationship between MV-induced mediator release and AKI. However, we describe a multi tude of potential eff ects on the kidney of several mediators that can be blocked rather specifi cally now in both animal models and humans. For example, anakinra, a synthetic IL-1RA, and infl iximab, a monoclonal antibody against TNF-α, have found their way into clinical practice.
Before targeting mediators to prevent AKI in patients, care must be taken to learn from past experiences. Attempting to alter the course of sepsis, numerous studies have targeted a variety of mediators in critically ill patients, mainly suff ering from sepsis. Th e rather disappointing results of these studies have led to insights into the possible mechanisms of failure in these studies. Several issues should be taken into account. Th e agent's biological activity, shown in vitro or in simple animal models, may not be replicable in humans. Dosage, timing and duration of the novel therapy are usually unknown. In most trials in critical care medicine the target population is heterogeneous, also including genetic polymorphisms. Th e complexity of mediator interdependency may also require the targeting of multiple mediators simultaneously or combined targeting of pro-and antiinfl ammatory mediators.

Conclusion
From a theoretical point of view, the systemic release of several mediators induced by MV may play an important role in the pathophysiology of AKI. However, evidence supporting this hypothesis or showing causal relationships is lacking for the studied mediators. Future studies should therefore not only focus on the release of mediators during MV and a possible relationship with AKI, but should also study in-depth the pathophysiology by which these mediators may contribute to AKI.