Granulocyte Colony-Stimulating Factor to Prevent the Progression of Systemic Nonresponsiveness in Systemic Inflammatory Response Syndrome and Sepsis

 THE SYSTEMIC inflammatory response syndrome (SIRS) constitutes the primary host response to a variety of severe clinical insults, such as trauma, burns, pancreatitis, or major surgical interventions.1 Distinct from SIRS, the term sepsis is applied to the clinical state when bacterial infections cause a systemic hyperinflammatory immune response.1 However, this exaggerated host inflammatory response may not lead to an efficient elimination of the infectious agent; rather, it may contribute to a state of immune deactivation that has been termed compensatory anti-inflammatory response syndrome (CARS).2 Concomitantly, CARS may increase susceptibility to secondary infections. The main pathways of the inflammatory circuit during SIRS, sepsis, and CARS are summarized in Fig 1.

The overall aim of this review is to discuss the current rationale for administering G-CSF to critically ill patients at risk of or with sepsis/SIRS to improve neutrophil function and to modulate the otherwise predestined release of inflammatory mediators. Thus, by direct and indirect effects, G-CSF may prevent the fatal course of sepsis in critically ill patients and promote recovery.

According to the general view, SIRS and sepsis are caused by trauma- or infection-induced overproduction of tumor necrosis factor-α (TNF-α) and interleukin-1 (IL-1).1 Severe sepsis with organ failure may develop when endotoxin (lipopolysaccharide [LPS]) from gram-negative bacteria and/or soluble products of gram-positive bacteria, mostly superantigens, lead to macrophage activation and production of macrophage and T-cell–derived cytokines, including TNF-α, IL-1, interferon-γ (IFN-γ), and IL-12. In turn, these cytokines may synergize with LPS to stimulate the release of nitric oxide (NO) via NOS-2 in a variety of cell types.3-8 

NO has been proposed as the direct mediator of the cardiovascular failure in sepsis9 and septic shock.10 Although NO is produced during sepsis from cytokine-regulated NOS-2, a strong correlation between NO production, organ failure, and an exaggerated proinflammatory response has only been documented in pediatric patients.11 In the majority of adult patients, a clear correlation is more difficult to prove, which may be due to suppression of TNF and IL-1 by anti-inflammatory cytokines, such as by IL-10, which downregulates NO synthase activity and the synthesis of proinflammatory mediators.12 NO synthase inhibition reduced TNF-stimulated IL-8 production in a human endothelial cell line, whereas exogenously added NO enhanced the release of IL-8.13 In addition, NO synthase inhibition blocked IL-8 and IL-6 production in LPS-stimulated human whole blood.14 Conversely, direct exposure of whole blood to NO caused a dose-dependent stimulation of IL-8 but had no effect on IL-6 release. Moreover, the hydroxyl radical scavenger dimethyl sulfoxide (DMSO) prevented NO induction of IL-8, suggesting the participation of the hydroxyl radical in the NO-induced IL-8 production.14 Thus, NO and reactive oxygen intermediates (ROI) may play an essential role in the cytokine pattern in critically ill patients with SIRS and sepsis.

In contrast, counterregulatory Th₂ type cytokines, such as IL-415 and transforming growth factor-β (TGF-β),4,16 have been shown to prevent NOS-2 induction. Importantly, the highly reactive peroxynitrite (ONOO⁻) is formed as a reaction product of NO plus ROI (Fig 1). ROI either result from xanthine oxidase activity17 or from neutrophils activated by bacteria or their soluble products. In addition to a number of other effects, peroxynitrite causes microvascular permeability and edema formation, leading to clinical signs of septic shock.18 As a potent oxidant, peroxynitrite reacts with lipids, thiols, DNA, and proteins and may constitute an inducer of apoptosis, as examplified in a study with vascular smooth muscle cells.7 19 

The fatal course of sepsis may be due to systemic immune dysregulation and immunosuppression. In septic mice, apoptosis has been shown to occur primarily in hematopoietic cells, such as T and B lymphocytes, as well as in parenchymal cells, such as bowel, lung, and, to a lesser extent, skeletal muscles and kidneys, but not in liver, brain, or heart.20 Thus, widespread lymphocyte depletion by apoptosis may contribute to lymphopenia observed in patients with sepsis and to decreased elimination of bacteria and bacterial products in sepsis.

Moreover, it has been shown that endothelial cell apoptosis can occur via an oxygen radical-dependent mechanism.21,22Experimental work by Ayala et al23,24 as well as Hotchkiss et al20 shows that systemic apoptosis of the immune system is the most marked feature in experimental animals dying of multiple organ failure during septic shock induced by massive bacterial infection, as already proposed by Bone.2 

In severely ill patients with sepsis, peripheral blood cell-derived TNF-α and IL-1β cytokine production after stimulation with LPS in vitro was transiently downregulated and restored within 72 hours in survivors, but there was no restoration in nonsurvivors.25Thus, persistent downregulation of proinflammatory cytokine production could be related to increased mortality in critically ill patients with sepsis. This downregulation may not result from the induction of tolerance, but through the selective removal of immune cells capable of producing these cytokines. Recent evidence suggests that, in addition, FAS-ligand (FasL) may mediate apoptosis in selected cell populations. Type 1 (Th₁) T helper lymphocytes, activated by bacteria, could be subsequently killed after FAS-ligand induction in the same or in a different T-cell population, a process that has been termed activation-induced cell death (AICD).26 AICD predominantly, but not exclusively, occurs in Th₁ cells.26,27Whereas T cells exposing FAS are highly sensitive to AICD, FAS-ligand–expressing cells are resistant.26,27 This may lead to a selective survival of Th₂ cells and a preponderance of Th₂ cells after proinflammatory response during SIRS and sepsis. In turn, predominant depletion of Th₁ cells by apoptosis may contribute to the development of CARS. The term CARS is based on the presence of an inactivating and compensatory response to the original inciting event, the proinflammatory response.2,28 This state of immunosuppression also results in increased susceptibility to infection,2 which may further contribute to an infection-induced fatal course of sepsis.

Agents contributing to the anti-inflammatory response include corticosteroids, prostaglandins, IL-4,15IL-10,29 IL-11, IL13, TGF-β,16 soluble receptors to TNF (sTNF-R), and receptor antagonists to IL-1 (IL-1ra)30-33 (Fig 1). The inhibitory Th₂ type cytokines IL-4, IL-10, and IL-13 downregulate monocyte activation states after encountering LPS.29,34 35 

To analyze the relative contribution of SIRS alone or in combination with infection, ie, sepsis, ex vivo cytokine production by neutrophils isolated from patients who had undergone cardiac surgery with cardiopulmonary bypass and from patients with sepsis were studied. These stressful conditions related to inflammation, independent of infection, rapidly dampened the reactivity of circulating neutrophils. The release of IL-8 by neutrophils in both groups of patients was significantly reduced whether activated by LPS or by heat-killed streptococci.36 LPS-induced mediators such as IL-10 may be responsible for the observed anergy in these patients. In sepsis, immune deactivation is also mediated by activation of TGF-β via proteases, such as plasmin.37 

Colony-stimulating factors, such as G-CSF and granulocyte-macrophage colony-stimulating factor (GM-CSF), appear to constitute a separate, unique class of mediators. These mediators not only trigger recruitment of progenitors from stem cells, but also modulate adhesion receptor expression as well as upregulate cell protective molecules. In mesenchymal cells such as fibroblasts and mesothelial cells, the G-CSF gene is clearly regulated in a coordinate manner with other genes encoding cytokines such as GM-CSF and IL-6,38-40 ie, these genes are all induced by the same stimuli with similar kinetics. After exposure of cytokines, such as TNF, induced by infections, G-CSF and GM-CSF are coinduced.38,39 In turn, both, G-CSF and GM-CSF may stimulate common effector pathways,41 42 while resulting effects of both cytokines may also differ significantly, as will be discussed in detail below.

In the clinical setting, three considerations are required for the successful treatment of patients with sepsis: (1) to stabilize blood pressure and maintain regional organ and tissue perfusion, (2) to modulate the hyperinflammatory immune response induced by specific (infections; sepsis) or nonspecific (ROI; SIRS) stimulators, and (3) to counteract systemic nonresponsiveness and anergy. The use of biological response modifiers for the treatment of patients with SIRS and sepsis is limited by the fact that the molecular nature of the inducer(s) is not precisely known and is likely not a single common denominator. In addition, differing clinical states have not yet been defined by immunological parameters, and treatment protocols are up to now not directed at a distinct immunological status of the patient.

These limitations become a challenge when biological response modifiers are considered as interventional therapies. Can any one single therapeutic approach help restore homeostasis in sepsis and SIRS regardless of its timing? Clearly, monotherapies such as inhibitors of TNF-α or IL-1 are limited in this regard. The putative goal for anti-inflammatory monotherapies has been to block or eliminate the inducing agent(s) or to modulate the proinflammatory cytokine cascade. The failure to demonstrate any benefit from such treatments43,44 may be due in part to the erroneous assumption that the pathogenesis of sepsis and SIRS in all patients is essentially one of persistent, uncontrolled inflammation driven by the overproduction of a single inflammatory mediator.2 

An alternative approach to the anti-inflammatory monotherapies has been to consider multipotent molecules, such as hematopoietic growth factors, which can address the various immunologic responses simultaneously in the septic patient. One of the hematopoietic growth factors, ie, G-CSF, plays a central role in proliferation, maturation, and functional activation of neutrophils.42,45-50 G-CSF has recently been shown to stimulate host immunity by increasing leukocyte number and by upregulating neutrophil function in postoperative/posttraumatic patients at risk of sepsis (Fig 2).51 IL-6 and G-CSF provide growth and survival signals to progenitors of the granulocyte lineage.52 G-CSF promoted mobilization and survival of CD34⁺ stem cells in the bone marrow53 and in the peripheral blood.54 This was due, at least in part, to suppression of apoptosis.53,54 Also, incubation of neutrophils from acquired immunodeficiency syndrome (AIDS) patients with G-CSF in vitro significantly decreased apoptosis.55 In patients with acute respiratory distress syndrome (ARDS), neutrophils from bronchoalveolar lavage fluid (BAL) showed only modest signs of apoptosis, presumably due to high amounts of G-CSF and GM-CSF as well as other cytokines like IL-6 in BAL.56 Thus, G-CSF may represent one of several survival factors for neutrophils to inhibit apoptosis during SIRS and sepsis and to preserve neutrophil function.

Besides these direct effects on granulocytes, G-CSF has been shown to exert direct and indirect effects on other immunocompetent cells. Direct effects are mediated by cells that express G-CSF receptors (G-CSFR). G-CSFR were clearly detected by flow cytometry on adult human peripheral granulocytes and monocytes, but not on lymphocytes.57 Administered prophylactically, direct and indirect effects of G-CSF might block the hyperinflammatory cytokine cascade mediated by proinflammatory cytokines before culminating in severe tissue damage. Moreover, these effects might help to restore immune dysfunction and dysbalance in critically ill patients. The addition of G-CSF promoted an anti-inflammatory response pattern in several usually proinflammatory test systems. For example, T cells of G-CSF–treated mice predominantly produced anti-inflammatory type-2 cytokines (IL-4 and IL-10).58 Moreover, G-CSF administration to human volunteers increased Th₂ type cytokines in LPS-stimulated ex vivo blood cultures.59Reduced Th₁ cytokine and increased Th₂ cytokine expression are expected to downmodulate NOS-2 activation, as schematically demonstrated in Fig 1.

As summarized in Fig 1, TNF-α can stimulate IL-1 secretion in tissue macrophages, and this cytokine in turn can activate IL-1ra production to counteract IL-1–mediated inflammatory cascades.60Exogenous G-CSF (dashed lines in Fig 1) rapidly stimulates IL-1ra production.59,61 62 The autocrine cytokine cascade of inflammatory mediators could thus be interrupted.

Moreover, G-CSF promotes IL-6 use via increased expression and shedding of the IL-6 receptor (IL-6R, p80), as shown independently in two patient populations.63,64 Indeed, IL-6 and G-CSF may act similarly by stimulating intracellular pathways, such as phosphorylation of STAT3 transcription factor.65-69 G-CSF and IL-6 cooperate and can even substitute for each other in recruitment of myeloid cells in G-CSFR and IL-6 knock out mouse models.52 G-CSF and IL-6 may act in a dual role to amplify inflammatory responses when the inciting pathogen is present within the host and to downmodulate the response when the pathogen has been eradicated.68 

Bacterial endotoxin stimulates macrophages to produce proinflammatory cytokines, such as TNF-α and IL-1, which in turn stimulate fibroblasts and endothelial cells to produce G-CSF.70 Dunn et al71 identified regulatory elements in the G-CSF promotor activated by NF-IL-6 and NF-κB, and Smith et al72 reported binding for the transcription factors PU.1 and C/EBP-α. Consequently, TNF-α synthesizing monocytes regulate G-CSF production by autocrine effects that are probably different from endothelial cells and fibroblasts. Inducers of G-CSF release may be distinct in different cell types, ie, macrophages, fibroblasts, and endothelial cells.

G-CSF serum concentrations have been reported to be less than 30 pg/mL in most healthy persons73 and rarely exceed concentrations greater than 170 pg/mL70,73-76(Table 1). Early during active infections, such as pneumonia, cholecystitis, or urinary tract infection, and sepsis, serum concentrations of G-CSF may increase rapidly to more than 50,000 pg/mL,70,74,77-80 but rapidly decrease due to clearance through binding and use. After an infectious episode, G-CSF serum concentrations returned to levels less than 200 pg/mL.70,74,79 80 

In recent years, G-CSF has been demonstrated to be effective and safe in reducing the incidence of infection in high-risk patients, ie, after myelosuppressive anticancer chemotherapy,45,81,82 as well as in neutropenic and agranulocytotic patients.82-84 Its effectiveness is due in part to G-CSF’s effects on increasing neutrophil count and functional receptors, such as G-CSFR as well as monomeric Fc receptors. Moreover, G-CSF might be beneficial in nonneutropenic patients with SIRS, infections, and sepsis with impaired neutrophil function. In these nonneutropenic patients, high concentrations of proinflammatory cytokines, such as TNF-α and IL-1, may cause accumulation of ROI in granulocytic effectors, thereby causing increased apoptosis of neutrophils,85 inefficient phagocytic function, and thus a state of immune dysfunction.

Secondary phagocyte defects and reductions in neutrophil numbers and functional impairment have been documented in a variety of clinical states. Patients with impaired neutrophil function are at increased risk to develop sepsis and to progress into septic shock and multiple organ failure.86,87 Preexisting immunosuppression due to the underlying disease has been reported in patients with diabetes,88,89 acute necrotizing pancreatitis,90 or alcoholism.91 Acquired immunosuppression and impairment of neutrophil function may result from previous medication (immunosuppressive and cytostatic drugs), operative trauma,92-94 blood transfusions,95 anesthetics, hypnotics, and sedatives96,97 as well as sympathomimetics.98 Migration to an infectious focus, ingestion, and killing of bacteria by neutrophils from critically ill patients with SIRS and sepsis has been reported to be significantly compromised.99-103 

However, boosting numbers and functions of granulocytes, macrophages, and lymphocytes is a prerequisite for withstanding microorganism invasion. Endogenous and exogenous G-CSF may be crucial for host defense in patients with SIRS and sepsis to avoid progression into severe sepsis with organ failure and septic shock by improving neutrophil functions and inhibiting neutrophil apoptosis.23,24,55 In addition to stimulating the proliferation and maturation of neutrophils,45 G-CSF enhances chemotaxis,46 phagocytic activity,47bactericidal function,47 respiratory burst,42,49 and antibody-dependent cellular cytotoxicity (ADCC)48 of isolated human neutrophils. Administration of G-CSF acts synergistically with antibiotics and improves outcome of severe infections.104-106 

G-CSF administration improved survival in several animal models of sepsis,105-108 even when applied therapeutically after the onset of sepsis105 (Fig 3). These animal data imply that stimulation of neutrophil function and viability represents a major goal for treating patients with ongoing infections.

There is now increasing evidence that an inappropriate endogenous G-CSF response may be associated with an adverse outcome to sepsis. For example, low serum G-CSF concentrations (0 to 125 pg/mL) on admission have been shown to be associated with fatal outcome in patients with acute bacterial infections.78 

In patients undergoing a successful response to infections, G-CSF serum concentrations should be high immediately and decrease consecutively, when G-CSF is used and neutrophil counts increase. In patients not responding successfully to infections, serum G-CSF concentrations will remain high, but neutrophil counts would not increase. Similar patterns are expected to occur when patients are treated with exogenous G-CSF.

In patients with bacteremia80 or at the time sepsis was diagnosed,79 endogenous G-CSF concentrations were found to be markedly increased in all patients, both survivors and nonsurvivors. However, concentrations significantly decreased with recovery in the survivors, but remained increased in the nonsurvivors. In the same study, the levels of G-CSF, IL-6, and, to some extent, TNF-α rapidly decreased in nonfatal bacteremic infections within the first 24 hours; however, concentrations remained increased or showed a more gradual decrease in the nonsurvivors within 18 to 96 hours.80 In a different study, application of exogenous rhG-CSF in 19 patients with granulocytopenia complicated by sepsis increased granulocyte counts and improved survival when G-CSF serum concentrations decreased.109 However, in the five patients with persistently high plasma G-CSF levels, leukocyte counts were not increased and the patients died.109 Thus, either early increased endogenous production or enhanced use of administered G-CSF (ie, G-CSF responsiveness) correlated with the patients’ successful outcome.

Thus, G-CSF responsiveness has to be monitored by leukocyte counts, G-CSF serum concentrations, and G-CSF–responsive gene products (see below) to ascertain successful treatment with exogenous G-CSF.

In a phase II study in patients at risk of or with sepsis, we successfully identified patients who could and could not respond to exogenous G-CSF administration. In responding patients, an increase in G-CSF serum concentrations after exogenous application of rhG-CSF was followed by an immediate decrease and a rapid increase of neutrophil counts and CD64 expression.61 In nonresponders, in which G-CSF substitution did not increase leukocyte counts, molecular mechanisms have to be defined. Impaired G-CSFR expression may contribute to nonresponsiveness. Whereas G-CSF induced increased G-CSFR expression,110 GM-CSF, TNF, LPS, f-met-leu-phe (fMLP), phorbol ester (TPA), and C5a were shown to downregulate the neutrophil G-CSFR in vitro.111 Binding assays using ¹²⁵I-labeled rG-CSF showed that the number of rG-CSF binding sites on the surface of neutrophil progenitor cells is regulated by cytoplasmic cAMP and protein kinase A.112 Rapid internalization of G-CSFRs111 may mask the amount of surface expression of G-CSFRs detected under G-CSF–responsive conditions. These conditions explain results showing reduced G-CSFRs in G-CSF–treated cell lines and neutrophils.113 Differential G-CSF responsiveness may ultimately explain the failure to demonstrate beneficial effects on nosocomial infections in a recently performed study in patients with acute traumatic brain injury or cerebral hemorrhage prophylactically treated with G-CSF.114 

Prophylactic administration with G-CSF appears to be superior to an endogenous immune response to fight infection in critically ill patients. During an infection, both G-CSF and GM-CSF are produced.38,39 Both cytokines induce leukocyte maturation and activation.41,42 In contrast to G-CSF, GM-CSF may cause serious side effects when applied exogenously and thus independently of G-CSF.45 115 

G-CSF inhibited leukotriene E₄(LTE₄) formation in humans in vivo116 and potently reduced the activity of NOS-2 in peritoneal macrophages.117 In contrast, GM-CSF upregulated NOS-2118-122 and enhanced LTB₄ generation of neutrophils.123-125 McDonald et al126 showed that IL-8 may be responsible for the synthesis of LTB₄ and platelet-activating factor after GM-CSF treatment. Thereby, the combined presence of IL-8 and of GM-CSF at inflammatory foci may contribute to the amplification of the inflammatory response. GM-CSF stimulated adhesion receptor expression, such as intercellular adhesion molecule-1 (ICAM-1).127 However, G-CSF downregulated ICAM-1 on IL-1–stimulated human endothelial cells.128Yong129 further showed that G-CSF, but not GM-CSF, promotes transendothelial migration of neutrophils. ROI released by adherent neutrophils may cause capillary leakage and lung toxicity, which occurred in patients with sepsis treated with GM-CSF.45,115 Massive IL-8 is considered to indicate the amount of ROI induced in vitro.13 14 

In contrast to GM-CSF, G-CSF is expected to downmodulate the proinflammatory immune response, which otherwise might result in severe tissue damage and organ failure. This rationale favors the administration of G-CSF to critically ill patients at risk of sepsis and before other proinflammatory cytokines are systemically present.

Functional antigens on neutrophils, such as CD64, CD32, CD16, and CD11b, defined by specific clusters of differentiation (CD), can be easily monitored by flow cytometry. rhG-CSF reconstituted and increased functional antigen CD64 (monomeric Fc receptor; FcγRI) expression in postoperative/posttraumatic patients at risk of or with sepsis,61 and the same increase of CD64 expression has been described on immature and mature neutrophils in normals and patients with malignancies130-132 (Table2A). Functionally, exogenous G-CSF enhanced ADCC in patients with advanced squamous cell carcinoma of the oral cavity and pharynx through FcγRI (CD64).130 Also, expression of FcγRII (CD32) increased when G-CSF was administered (Table2B).132,133 Functionally, G-CSF enabled neutrophils to kill hybridoma cell lines through FcγRII (CD32).134 FcγRIII (CD16) expression was upregulated by exogenous G-CSF in healthy volunteers75 and in patients with paroxysmal nocturnal hemoglobinuria,135 but did not change61,133 and even decreased131,132 in others (Table 2C). In human newborn infants with presumed bacterial sepsis, functional activation of neutrophils (C3bi = CD11b expression) was determined 24 hours after the application of 10 μg/kg·d of rhG-CSF.136 

Membrane-bound CD14 (mCD14) is a relevant surface antigen for effective endotoxin clearance. An enhanced expression of CD14 on neutrophils has been reported in healthy individuals after subcutaneous administration of G-CSF.131,137 Improved re-expression of mCD14 has been observed in G-CSF–treated postoperative/posttraumatic patients.64 The effects of G-CSF on mCD14 might explain a better LPS clearance.138 

The events leading to the upregulation of functional antigens by endogenous and exogenous G-CSF are explained in part by STAT1 and STAT3 phosphorylation65 and by transcription factors PU.1 and C/EBP-α72 activated through the G-CSF receptor. Both transcription factors play an important role in differentiation (PU.1) and induction of regulatory genes, induced by increased cytoplasmic cAMP112,139 and protein kinase A activation,112including the G-CSFR itself.112 These molecular events are relevant to understand specific as well as promiscous effects by G-CSF. Clearly, CD64 appears to be a highly valid marker to follow responsiveness to endogenous and exogenous G-CSF.

There is some concern that G-CSF–induced increases in neutrophil count and function might aggravate inflammatory responses and thus be harmful for the host. This concern is based on the functional properties of G-CSF to stimulate phospholipase A₂41 and to stimulate oxidative burst formation in granulocytes.42 On the other hand, these properties are most relevant to control infection and to attenuate or even prevent an otherwise progressive immune dysregulation. Direct effects (via G-CSF receptors on granulocytes and monocytes, expression of CD64, increase in neutrophil counts, and priming of neutrophil function) and indirect effects of G-CSF (clearance of endotoxin and bacteria and counterregulation of induction of proinflammatory cytokines) can prevent nonspecific tissue damage, especially by preserving endothelial cell function. Endothelial cell damage is expected to occur through systemically high TNF-α, IL-1, or IL-8 (Fig 1). Clearly, IL-8 is a cytokine released upon oxidative stress by almost every cell type.13,14 G-CSF treatment attenuates circulating IL-8.61 Further support for this concept of achieving less oxidative stress by G-CSF is provided in vivo by animal studies, when exogenous G-CSF improved rather than worsened organ function and survival in models using LPS- or live bacteria-induced proinflammatory response and disease.140Exogenous G-CSF increased microvascular flow and survival rate and prevented neutrophil-mediated tissue damage in fulminant intra-abdominal sepsis in rats, at least in part, by downmodulating the potent vasoconstrictor endothelin-1.105 Hemoconcentration and lactic acidosis were also attenuated, which is consistent with reduced endothelial damage and plasma leakage.105 

Table 3 summarizes the inflammatory cytokine response pattern resulting from application of exogenous G-CSF to animals.105,108 140-144 The relative contribution of direct (via G-CSF receptors on monocytes) and indirect effects (better control of infection, T cells) of G-CSF to this pattern has yet to be clarified. Clearly, G-CSF is a member of the inflammatory cytokine cascade and displays important regulatory properties by attenuating the proinflammatory cytokine response that would otherwise predominate.

Another secondary feature of exogenous G-CSF administration is its effects on T cells. In mice, G-CSF has been shown to polarize T-cell subpopulations towards an increased anti-inflammatory type-2 cytokine response (characterized by the cytokines IL-4 and IL-10) and a reduced proinflammatory type-1 cytokine production (IL-2 and IFN-γ)58 (Table 4). In a similar pattern, G-CSF administration altered serum concentrations of cytokines in healthy volunteers145 and ex vivo cytokine production of whole blood in response to gram-positive and gram-negative bacterial products,59,145 respectively, in that more anti-inflammatory and less proinflammatory cytokines were released (Table 4). Although, G-CSF administration to healthy volunteers caused a reproducible increase of proinflammatory mediators, such as TNF-α, after ex vivo and in vivo stimulation with LPS, the increases of counterregulatory and anti-inflammatory cytokines, such as the soluble p55 and p75 TNF receptors (sTNF-Rs) and IL-1ra, were significantly higher.62,145 Similarly, Hartung et al59 found more anti-inflammatory cytokines, such as IL-1ra, sTNF-R-type I and -type II, and IL-10, but less proinflammatory cytokines, such as TNF and IFN-γ, in blood cultures stimulated with bacterial products from volunteers treated with 480 μg G-CSF applied subcutaneously.

G-CSF exerts differential effects on cytokine release in human endotoxemia when administered either intravenously forn 2 hours or subcutaneously for 24 hours before endotoxin.62 When administered 2 hours before endotoxin, G-CSF actually augmented an LPS-induced inflammatory cytokine response, whereas when administered 24 hours before LPS challenge, G-CSF attenuated the LPS-induced proinflammatory state and caused significant downregulation of inflammation (Table 4).62 A direct explanation implies the involvement of a G-CSF–induced transcription of gene product(s) that functions to regulate the proinflammatory effects induced by endotoxin. If G-CSF causes modulation of LPS-receptor expression, such as CD14, the anti-inflammatory gene product(s) may be either involved in endotoxin removal or in neutralizing oxidative stress.62Further experimental work is necessary to understand the regulatory pathways induced by G-CSF in the absence of a proinflammatory stimulus. Results are fundamental to work out a prophylactic G-CSF treatment schedule applicable to patients at risk of sepsis.

Adhesion molecules, such as selectins, regulate the primary binding of neutrophils to the endothelium and β₂-integrins are necessary for subsequent transendothelial migration.146-149G-CSF may decrease selectin expression and diminish binding to and enhance transmigration of neutrophils through the endothelium. Therefore, G-CSF may prevent endothelial damage by avoiding adhesion of activated neutrophils to inflamed vascular endothelium.

Downregulation of L-selectin by G-CSF may be beneficial during distinct pathological inflammatory responses, which are associated with the adhesion of hyperactivated neutrophils to the endothelium. L-selectin–deficient mice were resistant to LPS-induced toxic shock and death, because neutrophil, lymphocyte, and monocyte migration to the peritoneum in response to an inflammatory stimulus was significantly inhibited.150 In healthy volunteers, exogenous G-CSF reversibly downregulated surface expression of L-selectin (synonymous [syn.]: leukocyte adhesion molecule-1 [LAM-1]) on human neutrophils151 and a specific surface metalloproteinase, called L-selectin sheddase, may be involved.152 In 11 postoperative/posttraumatic patients at risk of or with sepsis and treated with rhG-CSF, L-selectin expression on granulocytes was downregulated within 24 to 48 hours.61 

Increased plasma concentrations of soluble E-selectin, indicating endothelial damage, were closely associated with multiple organ dysfunction and death in patients with SIRS.153 Patient survival was zero when soluble E-selectin levels were greater than 30 U/mL.153 Exogenous G-CSF may have been protective to the endothelium, because in postoperative/posttraumatic patients at risk of or with sepsis, soluble E-selectin showed a sharp increase when rhG-CSF was reduced from 1 to 0.5 μg/kg·d.154 

The intercellular adhesion molecule (ICAM-1; CD54) is expressed on the surface of inflamed endothelial cells and plays an important role in specific binding for transmigration of activated granulocytes.147-149 An increase in soluble ICAM (sICAM) may also be indicative of endothelial damage. In this context, it has been reported that sICAM-1 serum concentrations were increased in adult patients with sepsis and corresponded with severity of disease, with subsequent organ failure, and possibly also with outcome.155 In postoperative/posttraumatic patients at risk of or with sepsis,61 sICAM concentrations were relatively stable during the infusion of rhG-CSF at 1 μg/kg·d; however, levels increased when rhG-CSF was tapered to 0.5 μg/kg·d and after cessation of rhG-CSF. Nethertheless, extrapolating sICAM concentrations to endothelial damage remains controversial, because ICAM-1 is also expressed and can be shed from monocytes.156 

Despite upregulation of CD11b/18, G-CSF had no effect on neutrophil adhesion but was a powerful stimulator of transmigration.129 In whole blood cultures, a time- and G-CSF concentration-dependent upregulation of the expression of CD11b has been observed.75 In healthy volunteers, the nadir in circulating neutrophils occurred 30 minutes postinjection of 300 μg G-CSF subcutaneously and coincided with significant upregulation of the expression of CD11b on circulating neutrophils.75 This effect was short-lasting in that CD11b was maximally expressed at 2 hours postinjection and returned to baseline levels at 4 hours postinjection.75 

Taken together, diminished but not abolished binding of preactivated neutrophils to endothelial cells and enhanced transmigration, induced by exogenous rhG-CSF, might effectively protect the endothelium from injury by hyperactivated neutrophils.

The encouraging results in nonneutropenic patients have supported expanding the application of G-CSF into nonneutropenic, critically ill patients with secondary functional impairment of neutrophils.

G-CSF should benefit the course of disease in patients with pneumonia and subsequent organ dysfunction (Table 5). The neutrophil has been strongly implicated in the pathogenesis of inflammatory lung injury,50 and there has been theoretical concern that G-CSF and neutrophil activation might exacerbate lung injury. Exogenous G-CSF clearly worsened lung injury and survival when applied during very severe pulmonary infection in rats.50In other studies in which lung injury was exacerbated in G-CSF–treated animals and humans, a beneficial survival effect remained significant (Table 5).50,157,158 In ethanol-treated rats with experimentally induced pneumonia, all control animals, but less than 10 % of the G-CSF–treated rats died.159 Also, exogenous G-CSF improved survival in models in which pneumonia was induced in aminals by various microorganisms, such as Klebsiella pneumoniae,159 Streptococcus pneumoniae,160,161and Pseudomonas aeruginosa.162 In a canine model of peritonitis and septic shock, G-CSF application did not aggravate sepsis-related pulmonary dysfunction despite increases in circulating as well as lung lavage neutrophils.141 In mice, exogenous rhG-CSF has been shown to decrease LPS-induced pulmonary edema and alveolar capillary leakage.140 In a canine model of peritonitis and septic shock, exogenous G-CSF did not aggravate sepsis-related pulmonary dysfunction despite increased circulating as well as lung lavage neutrophils.141 

In human volunteers, rhG-CSF pretreatment prevented the accumulation of neutrophils in the lung during the first 2 hours after endotoxin administration.62 This prevention occurred in the presence of increased CD11b and CD18 expression on the neutrophils, ie, under otherwise proadhesive conditions.62 In neutropenic patients, G-CSF application predominantly resulted in an improvement in pulmonary function.91,157,158,163 G-CSF has also been proven to benefit nonneutropenic patients with pneumonia and subsequent organ dysfunction. In a study involving 756 hospitalized patients with community-acquired pneumonia, rhG-CSF accelerated clearance of pulmonary infiltrates and reduced the occurrence of serious complications: empyema, ARDS, disseminated intravascular coagulation, and septic shock.164 The frequency of nosocomial pneumonia was higher in patients with acute traumatic brain injury or cerebral hemorrhage prophylactically receiving 300 μg/d rhG-CSF (4/14) than in the controls (2/17); however, the difference with respect to the control group was not statistically significant.114 For several reasons, this study was underpowered to detect a difference and thus not ideal to evaluate a potential benefit or adverse effect from the prophylactic application of G-CSF in patients with high risk for nosocomial infections. In a different study with a small number of postoperative/posttraumatic patients at risk of or with sepsis, the application of rhG-CSF did not affect pulmonary function.165 Combined with the molecular events discussed above, G-CSF is unlikely to promote nonspecific adhesion of activated neutrophils to inflamed tissues and thus cause tissue damage as long as other proinflammatory mediators such as TNF-α, IL-1, and GM-CSF are low or absent.

The present studies in humans and animals stress the importance of the appropriate timing, dosage, and patient selection for a prophylactic treatment protocol with G-CSF in pneumonia. G-CSF prophylactically administered after hemorrhage in mice improved survival from pneumonia due to Pseudomonas aeruginosa; however, the protective effect was highly dependent on the dosing schedule used.162 In a nonneutropenic infection model of Streptococcus pneumoniaepulmonary infection, rhG-CSF improved lung clearance in both splenectomized and sham-operated mice compared with controls; however, rhG-CSF improved survival in the splenectomized mice but not in the sham-operated mice.160 In ethanol-fed rats, G-CSF was unable to provide protection against fatal Pneumococcal pneumonia despite increasing the numbers of circulating neutrophils.161 

Recently, a dramatic increase in mortality has been reported when mice were pretreated with G-CSF before induction of pneumonia withKlebsiella pneumoniae.166 In vitro, G-CSF increased capsular polysaccharide (a bacterial virulence factor) production ofKlebsiella pneumoniae, resulting in impaired phagocytic uptake and killing by neutrophils.166 Hemorrhagic shock in rats was accompanied by acute lung injury with neutrophil infiltration and an increase of G-CSF mRNA levels in the lung produced mainly by bronchial epithelial cells.167 G-CSF instillation into rat lungs mediated neutrophil recruitment, pulmonary edema, and hypoxia, indicating that local production of G-CSF may be involved in lung damage and ARDS.168 

Thus, depending on certain circumstances, application of G-CSF for prophylaxis and therapy of pneumonia may be beneficial or detrimental. Appropriate dosage, timing, and indication are not yet clearly defined to gain maximal beneficial effects with minimal side effects in patients. In summary, recent studies114,164,166 168indicate the need for additional controlled studies to define the role of G-CSF in community-aquired and nosocomial pneumonia.

At present, exogenous G-CSF is widely used and has been proven to be effective and safe in reducing the incidence of infection and sepsis in immunocompromised patients with nonmyeloid tumors after myelosuppressive anticancer chemotherapy45,81,82 as well as in neutropenic and agranulocytotic patients.82-84 

In human newborn infants with presumed bacterial sepsis, application of G-CSF significantly increased C3bi (CD11b) expression of their neutrophils, indicating functional activation.136 In nonneutropenic, posttraumatic/postoperative patients with a high risk of sepsis, none of the patients treated with rhG-CSF developed sepsis; however, three patients in the control group did.51 In fulminant intra-abdominal sepsis in rats, mortality decreased from 92% to 46% and to 42%, respectively, with the first dose of G-CSF at sepsis induction or 4 hours postinduction, ie, during manifested sepsis, respectively (Fig 3).169 Based on animal studies, G-CSF should be effective not only in prophylaxis, but also in treatment of sepsis.

In future use, G-CSF may be effective in nonneutropenic critically ill patients suffering from yet unrecognized impairment of neutrophil function and tissue injury. Beneficial effects of exogenous G-CSF have been observed on the resolution of infection, especially pneumonia, and subsequent sepsis, multiple organ dysfunction, and septic shock. These effects of exogenous G-CSF result directly from a better control of infection and indirectly from an amelioration of an overwhelming, damaging, proinflammatory response.

Ischemia/reperfusion injury is also a frequent consequence of surgical injury, hemorrhagic shock, and trauma. Improved endotoxin elimination by exogenous G-CSF138 may contribute to the recently described endothelial cell protection by the immediate induction of IL-1ra. This latter property of G-CSF extends the therapeutic potential to ischemia/reperfusion injury. Short, clinically relevant visceral ischemia produced TNF-α– and IL-1–dependent lung injury in mice.170 Recently, in rats with visceral ischemia/reperfusion injury, treatment with rhG-CSF increased survival, reduced serum TNF-α induction, reduced myeloperoxidase activity in the ileum and in the lung, and improved mean arterial blood pressure.117 Such results may be explained by our recent observation that G-CSF stimulated hemoxygenase-1 (HO-1) expression in microvascular endothelial cells in vitro (unpublished observation). If HO-1 induction by G-CSF is a general phenomenon in endothelial cells, then this may represent an important response against oxidative damage.

G-CSF has been demonstrated to alleviate neutropenia in patients with advanced AIDS or AIDS-related complex receiving zidovudine.83 In addition, the prophylactic use of G-CSF may prevent endotoxin-induced shock in AIDS patients. The incidence of infection with gram-negative bacteria is increased in AIDS patients who suffer from accelerated leukocyte apoptosis and defective neutrophil functions.55,171 Accelerated apoptosis contributes to impaired neutrophil function in AIDS patients.55 However, incubation with G-CSF in vitro significantly decreased apoptosis in neutrophils of these patients.55 In addition, repeated pretreatment with rhG-CSF successfully protected mice with retrovirus-induced murine acquired immunodeficiency syndrome (MAIDS) from hypersensitivity to LPS-induced lethal shock, and this protective effect coincided with the suppression of IFN-γ production.144 

Application of rhG-CSF may also find widespread use in critically ill patients, because no serious side effects have been observed in multiple therapeutic trials in immunocompromised, agranulocytotic, neutropenic, and nonneutropenic patients.45,51,61,82,136,164,165 G-CSF is well-tolerated, and the most common side effects, bone pain and occasionally reported skin rash, appear to be dose-dependent.45,82 Transgenic mice that overexpressed G-CSF had a 100-fold increase in G-CSF serum concentrations (1,041 ± 242 pg/mL in sera) and peripheral neutrophils. These G-CSF–overexpressing transgenic mice developed osteoporosis because of increased osteoclastic activity. Thus, high G-CSF serum concentrations could have a negative influence on bone homeostasis in vivo when active over extended periods of time.172 

Animal studies suggest that prolonged therapy with G-CSF may also have anti-inflammatory effects in colitis. Inflammatory bowel disease is associated with mucosal neutrophil recruitment and activation, mediated in part by arachidonic acid metabolites. G-CSF attenuates the immune response to sepsis and ameliorates glycogen storage disease Ib-related colitis. In immune complex colitis in rabbits, treatment with rhG-CSF resulted in a marked decrease of proinflammatory mediators, such as LTB₄ and thromboxane B₂ (TXB₂), but mucosal generation of the protective prostaglandin E₂(PGE₂) was preserved.173 

In critically ill patients with a secondary functional impairment of host defense, G-CSF improved leukocyte recruitment and, in addition, appears to activate a number of cell protective mechanisms awaiting more detailed investigations.

The present data support the hypothesis that G-CSF treatment may be extended to a group of critically ill patients with secondary impairment of neutrophil function occurring as a result of extensive tissue damage, oxidative stress, and/or ischemia/reperfusion. G-CSF both augments the replication and function of neutrophils. G-CSF also stimulates the elimination of endotoxin and attenuates the hyperinflammatory state by increasing IL-1ra and by intermittently modulating the adhesion receptor L-selectin. These events would counteract the progression of sepsis and, presumably, the incidence and severity of multiple organ dysfunction. Refinement in the use of G-CSF, either prophylactically or in states with acute infection, reperfusion, or severe tissue injury relies on the experimental as well as clinical evidence that G-CSF stimulates neutrophil function and concomitantly attenuates an uncontrolled inflammatory response. However, the successful administration of G-CSF to nonneutropenic patients with impairment of neutrophil function will strongly depend on optimal timing and dosage. The prophylactic administration of G-CSF to surgical patients with low G-CSF serum concentrations will result in appropriate serum concentrations of about 1,000 pg/mL G-CSF in responding patients when administered at about 1 μg/kg·d.51,61,165 These doses are associated with beneficial effects on neutrophil function in the absence of granulocytosis greater than 50,000/μL.51,61 165 

An ongoing inflammation is possibly also targeted by G-CSF with beneficial effects if the G-CSF administration is associated with its efficient use. Evidence of use includes functional activation (CD64 expression, phagocytic activity, and less apoptosis), decreased L-selectin expression on neutrophils, and a concomitant increase of serum IL-1ra as well as a decrease of serum proinflammatory cytokine concentrations (such as IL-8). Thus, the definition of subgroups of patients sensitive to G-CSF treatment, which by now includes cases with pneumonia, diabetes, acute necrotizing pancreatitis, alcoholism, AIDS, and ischemia/reperfusion injury, may further extend our understanding of immunomodulatory effects by this hematopoietic growth factor.

Additional controlled studies with critically ill patient populations at a high risk to encounter severe infections and sepsis are necessary to further define and use the regulatory properties of G-CSF in a window of effective treatment protocols.