American Journal of Respiratory and Critical Care Medicine

Granulocyte-macrophage colony-stimulating factor (GM-CSF) stimulates hemopoiesis and effector functions of granulocytes and macrophages and is involved in pulmonary surfactant homeostasis. We investigated whether GM-CSF therapy improved clinically diagnosed severe sepsis and respiratory dysfunction in critically ill patients. This randomized, double-blind, placebo-controlled phase II study added low-dose (3 mcg/kg) intravenous recombinant human GM-CSF daily for 5 days to conventional therapy in 10 patients, with a further eight patients receiving placebo. GM-CSF–treated patients showed improvement in PaO2/FiO2 over 5 days (p = 0.02) and increased peripheral blood neutrophils (p = 0.08), whereas alveolar neutrophils decreased (p = 0.02). GM-CSF therapy was not associated with decreased 30-day survival or with increased acute respiratory distress syndrome or extrapulmonary organ dysfunction. GM-CSF therapy was associated with increased blood granulocyte superoxide production and restoration or preservation of blood and alveolar leukocyte phagocytic function. We conclude that low-dose GM-CSF was associated with improved gas exchange without pulmonary neutrophil infiltration, despite functional activation of both circulating neutrophils and pulmonary phagocytes. In addition, GM-CSF therapy was not associated with worsened acute respiratory distress syndrome or the multiple organ dysfunction syndrome, suggesting a homeostatic role for GM-CSF in sepsis-related pulmonary dysfunction.

Studies of antiinflammatory and immunomodulatory therapies for severe sepsis and septic shock (1) have suggested that patients suffering from sepsis may be considered immunocompromised hosts. Some of the acquired immunologic abnormalities that follow sepsis, trauma, or major surgery have been reversed with interferon-γ (2) and in vitro granulocyte-macrophage colony-stimulating factor (GM-CSF) (3, 4), although the potential clinical utility of these approaches remains unknown.

Both GM-CSF and granulocyte colony-stimulating factor (G-CSF) stimulate adhesion molecule expression, phagocytosis, and oxidative functions of host defense neutrophils, with GM-CSF also having similar actions on monocytes and macrophages (5). A lack of GM-CSF in mutant mice leads to disturbed pulmonary surfactant homeostasis with an alveolar proteinosis-like disease (6) and a marked propensity to develop pulmonary (7) and soft tissue infections that may exceed the rate of infection in G-CSF–deficient mice (8).

GM-CSF administration improves pulmonary gas exchange in some patients with acquired alveolar proteinosis (9), accelerates the healing of cutaneous ulcers (10), and restores impaired human neutrophil phagocytic function in vitro following trauma (11).

This phase II study in critically ill patients with severe sepsis and respiratory dysfunction tested the hypothesis that a low-dose daily infusion of GM-CSF enhances leukocyte function in critically ill patients with severe sepsis and sepsis-related pulmonary dysfunction without exacerbation of pulmonary or other organ dysfunction.

Subjects and Study Design

During a two-year period, consent for study inclusion was available for 18 patients. These subjects represented approximately one-third of patients with severe sepsis treated in the intensive care unit (ICU) of The Royal Melbourne Hospital during this time. The study was a randomized, double-blind, placebo (PL)-controlled phase II trial of five consecutive daily infusions of low-dose (3 μg/kg body weight) recombinant human GM-CSF (Molgramostim; Schering-Plough, Baulkham Hills, New South Wales, Australia) or PL in addition to conventional intensive care management of sepsis. In the 24 hours before entry, each patient had severe sepsis (12), consisting of two or more systemic inflammatory response syndrome criteria, including T more than 38°C or less than 36°C caused by clinically suspected infection and sepsis-related pulmonary organ dysfunction (defined in this trial as PaO2 of less than 60 mm Hg breathing room air or an a/A ratio of less than 0.6 [a PaO2/FiO2 ratio of less than 287] with a pulmonary infiltrate on chest radiograph).

Patients were excluded from participation if they were less than 15 years of age, had any active malignancy, required chemotherapy or ongoing immunosuppressive therapy, had received corticosteroids within 4 weeks before entry, or if pregnancy was not excluded.

The patient's usual treating physicians continued all relevant conventional therapy. Mechanical ventilation was provided with Puritan Bennett 7200 series ventilators (Nellcor Puritan Bennett, Pleasanton, CA). Consent was obtained from the subject's next of kin or legal representative. The investigation protocol was approved by The Royal Melbourne Hospital Board of Medical Research/Scientific Committee and by its Ethics Committee on Research.

Randomized, packaged, and coded placebo (PL) and lyophilized GM-CSF (GM) were presupplied by Schering-Plough Corporation for two groups of 10 patients in identical vials containing 400 mcg of GM-CSF or PL. Each dose of lyophilized GM or PL was reconstituted with 1 ml of sterile water and administered diluted in 50 ml of 0.9% saline via a central venous catheter over a period of 5 hours. The randomization code (GM or PL therapy corresponding to patient numbers 1 through 20) was not revealed to the clinical researchers until all patient data collection had been completed.

The primary trial outcome was patient survival to 30 days after trial entry. Secondary outcomes, assessed at 5 days, were the effect of GM-CSF on oxygenation (using the most favorable PaO2/FiO2 data from each day), the occurrence of the acute respiratory distress syndrome (ARDS) (13), the degree of organ dysfunction measured by the sepsis-related organ failure assessment score (14), and the number and functional status of blood and bronchoalveolar leukocytes.

Cell Samples and Measurements

Before the first and after the fifth dose of GM/PL, cells were collected from peripheral blood (neutrophils) and bronchoalveolar lavage (BAL) (mixed neutrophils and alveolar macrophages) and processed using methods similar to those described recently (15, 16).

Procedure and location of BAL.

BAL was performed on entry in all 18 patients, and paired samples on Day 5 were collected in all but three cases (case 3 GM and 6 PL were extubated before Day 5, and case 9 PL was critically ill receiving 100% O2). Using a fiberoptic bronchoscope, 4 × 50 ml aliquots of 0.9% sterile saline were infused into one segment and removed by gentle suction into plastic traps and transported quickly to the laboratory on ice. Median volume recovery was 57% on entry and 58% on Day 5. Paired BAL was performed in the same pulmonary segment on entry and Day 5. The right middle lobe was the preferred sampling site (11 of 15 [73%]), but where this was technically unsatisfactory, the right upper lobe (3 of 15 [20%] or a lower lobe segment (1 of 15 [7%]) was used. The same physician performed all BAL procedures.

Preparation of cells.

Whole peripheral blood was collected into ethylenediaminetetraacetic acid anticoagulant tubes at room temperature, and total white blood cell suspensions (“buffy coat”) were prepared by centrifugation at 300 × g for 10 minutes. Residual red cells were removed with 0.84% ammonium chloride lysis buffer (Sigma, Castle Hill, Australia) followed by washing in phosphate-buffered saline (Unipath; Basingstoke, Hampshire, UK).

Laboratory analysis of BAL cells was performed after exclusion of the first (bronchial) return aliquot and pooling of the return from the other three (alveolar) aliquots. Alveolar lavage fluid was not filtered through gauze, but macroscopic mucoid debris was removed by pipette, and then the fluid was centrifuged at 100 × g for 10 minutes to sediment BAL cells. Subsequently, these cells were resuspended in phosphate-buffered saline and were counted by a hemocytometer. Viability was assessed by trypan blue exclusion (median 91% on both days), and differential cell counting was performed in cytocentrifuge preparations using approximately 1 × 105 cells per slide (Cytospin 2; Shandon, Runcorn, Cheshire, UK).

Assessment of blood neutrophil respiratory burst and phagocytic activity used cells isolated by dextran sedimentation (dextran T500; Pharmacia, Piscataway, NJ) and density gradient centrifugation through Lymphoprep (Nyegaard and Co., Oslo, Norway) (17).


Phagocytosis of zymosan was determined by measurement of cellular 125I incorporation, as described (18). Briefly, 106 cell aliquots were incubated in a 96-well plate (Falcon; Becton Dickinson Labware, Bedford, MA) for 1 hour with 50 μL Tyrode buffer containing 125I and 80 mg/ml (adjusted to an optical density of 1.6 at λ = 540 nm) of the yeast cell wall product zymosan (Sigma Chemical Co., St. Louis, MO). Zymosan particles were phagocytosed by the neutrophils and/or macrophages, which consequently underwent radioiodination catalyzed by endogenous myeloperoxidase. Cells were then harvested onto glass fiber filters (Filtermate 196; Packard Instruments Corp., Downers Grove, IL), and the 125I retained on the filter (macromolecule incorporated) was measured using a Top Count scintillation counter (Packard Instruments Corp.). The amount of 125I retained on the filter was expressed as the percentage of total radioactivity in the incubation.

Superoxide production.

Superoxide anion (O2) generation, stimulated in vitro by either 100 nM formylmethionyl-leucylphenylalanine (Sigma) or 400 μg/ml zymosan (Sigma), was measured colorimetrically by monitoring cytochrome c absorbance as described previously (17).

Statistical Analysis

Appropriate data transformations (logistic for proportions and logarithmic for cell counts) with subtraction of background readings were performed as required (19). Changes in continuous variables over time were compared with paired t tests, whereas overall group and treatment effects were compared with repeated-measures analysis of variance. Distribution free analyses (Mann-Whitney and Wilcoxon) were also used, and differences between proportions were assessed by chi-square or Fisher exact tests, as appropriate. Survival over time in each treatment group was compared with the log-rank test. All patients were included in the analyses, except where missing data necessitated exclusions; p ⩽ 0.05 was considered statistically significant.


Ten patients received GM and eight PL, with the primary diagnoses leading to intensive care admission being community-acquired pneumonia (two PL, three GM), alcohol abuse and pneumonia (two PL, one GM), burns (two PL, one GM), chest trauma (three GM), peritonitis (one PL, one GM), and ruptured esophagus (one PL, one GM). The median duration of mechanical ventilation in ICU before trial drug commencement was similar between groups (PL 4.5 days and GM 3 days, p = 0.53). On trial entry, the study groups were well matched, except that the GM group was significantly older (Table 1)

TABLE 1. Characteristics of patients on trial entry

 n = 8

 n = 10

Male, n (%)6 (75) 7 (70)1.00*
Age, years, median (range)46.5 (37 to 60) 62 (32 to 73)0.02
Weight, kg, median (range)75 (55 to 85) 75 (60 to 100)0.86
Leukocytes, ×109/L, mean ± SD (range)14.0 ± 5.7 (8.5 to 24.8)11.6 ± 5.7 (3.9 to 22.0)0.38
Neutrophils, ×109/L, mean ± SD (range)11.7 ± 5.5 (6.7 to 21.5)8.7 ± 4.4 (2.9 to 18.0)0.22
PaO2/FiO2,§ mean ± SD (range)166 ± 44 (115 to 250)160 ± 38 (100 to 228)0.77
APACHE II score, median (range)10 (5 to 22) 12 (3 to 20)0.93
ARDS, proportion (%)37.5401.00*
SOFA score, median (range)
6 (4 to 11)
 7 (5 to 12)

*Exact test, two-tailed.

Mann-Whitney test.

Unpaired t test.

§Best recorded gas exchange in the 24-hour period before trial entry.

Definition of abbreviations: ARDS = acute respiratory distress syndrome; GM-CSF = granulocyte-macrophage colony-stimulating factor; SD = standard deviation; SOFA = sepsis-related organ failure.

. All patients completed 5 days of therapy in the ICU, except for one patient with pneumonia, who received PL and left the ICU on the fourth trial day. There was no apparent difference in survival rate at 30 days (PL 75% and GM 80%, p = 0.84) or at discharge from hospital (PL 63% and GM 50%, p = 0.66), nor was there a difference in the median duration of ICU admission (PL 17.5 days and GM 21.5 days, p = 0.59).

All patients were receiving antibiotics on trial entry, and organisms isolated from blood collected before trial entry were Streptococcus pneumoniae (one PL), Klebsiella pneumoniae (one GM), and coagulase negative Staphylococcus sp. (one GM). Samples of BAL taken on trial entry in 13 patients grew no organisms. In the remaining patients, the most commonly isolated organism was Staphylococcus aureus (mostly methicillin resistant), either alone (one GM) or mixed with Pseudomonas aeruginosa (one PL) or Citrobacter freundii (one PL, one GM). One patient (PL) had pneumonia caused by Legionella pneumophila. Of the 15 pulmonary samples collected after 5 days, most were microbiologically sterile, with isolation of MRSE and Candida albicans in one PL case each, whereas methicillin-resistant Staphylococcus aureus (MRSA) persisted in the lavage of one GM patient. One patient had a documented abnormal blood culture after trial commencement, this being MRSA on Day 10 in a GM patient.

Pulmonary Gas Exchange and Acute Respiratory Distress Syndrome

The addition of GM to treatment was associated with an improvement in oxygenation over 5 days, whereas there was no significant change in gas exchange in patients who received PL (Figure 1

and Table 2)

TABLE 2. Effect of 5 days of treatment with gm-csf

Placebo (n = 8)

GM-CSF (n = 10)

Day 0
Day 5
Day 0
Day 5
PaO2/FiO2, mean ± SD (range)157 ± 49 (115 to 250)158 ± 50 (89 to 240)0.98*136 ± 52 (63 to 228)185 ± 53 (110 to 277)0.02*
ARDS, proportion (%)37.562.51.040200.25
SOFA score, median (range)6 (4 to 11)7 (4 to 16)0.397 (5 to 12)6.5 (3 to 14)0.80
Blood neutrophils, ×109/L,
   mean ± SD (range)
11.7 ± 5.5 (6.7 to 21.5)
11.2 ± 10.2 (0.7 to 32.8)
8.7 ± 4.4 (2.9 to 18.0)
13.1 ± 8.3 (3.7 to 28.6)

*Paired t tests using log-transformed results.

McNemar symmetry chi-square test.

Wilcoxon sign rank test.

Definition of abbreviations: ARDS = acute respiratory distress syndrome; GM-CSF = granulocyte-macrophage colony-stimulating factor; SOFA = sepsis-related organ failure.

. The maximum difference between mean “best” PaO2/FiO2 ratios of GM and PL groups (62.1 mm Hg) occurred near the completion of GM-CSF therapy, 96 hours after trial entry. The relative improvement in gas exchange associated with GM-CSF therapy compared with PL was found to persist to approximately the same degree following cessation of therapy for at least another 5 days (Figure 1).

Ventilatory management was similar between the two treatment groups during this time. PaCO2 amounts were evaluated on Day 0 and Day 5 at the times when the PaO2/FiO2 ratio was calculated. The mean PaCO2 amounts on the day of trial entry were very similar between PL and GM groups (45.9 mm Hg versus 44.4, p = 0.55). On Day 5 likewise, there were no differences in PCO2 amounts that would have accounted for the observed differences in PaO2 (43.6 versus 45.3 mm Hg, p = 0.69). PL and GM treatment groups were also well matched with respect to maximum continuous positive airway pressure used during mechanical ventilation, both from entry to 5 days (mean, 10.5 cm H2O both groups, p = 1.0) and up to 10 days (mean PL, 10.9 cm H2O; GM, 10.5 cm H2O, p = 0.79). Also, the ventilator-delivered tidal volume per kg of patient weight for each day from trial entry until Day 5 was similar between trial groups. The overall median of each daily maximum ventilator tidal volume received by PL patients was 10.9 ml/kg, whereas GM patients received a median of 10.5 ml/kg (p = 0.21).

Likewise, fluid management was similar between both treatment groups. Central venous pressure measurements were available in all GM and six of eight PL patients throughout the study. The median central venous pressure amounts, expressed in mm Hg, were comparable between groups on trial entry (PL 7 and GM 9.5, p = 0.17) and after 5 days (PL 8.5 and GM 6, p = 0.55). Pulmonary artery catheters were not mandatory but were present in nine patients (3 PL and 6 GM on trial entry; 4 PL and 5 GM on Day 5). Pulmonary artery wedge pressure (PAWP) pressures, expressed in mm Hg as the median of all measurements taken on trial entry and Day 5, respectively, were 12, 14 (PL) and 15, 16 (GM). PL patients required slightly more inotrope therapy and, in parallel, also received slightly more intravenous fluid through the 5-day trial period, with both these differences failing to reach statistical significance. The median daily positive balance was PL 800 ml versus GM 437 ml (p = 0.13). During the study period, infusion of at least one of four inotropes (norepinephrine, epinephrine, dopamine, or dobutamine) occurred for a median of 2 days (range, 0 to 5 days) in PL patients, whereas GM patients received a median of 0.5 days (range, 0 to 5 days) (p = 0.46). The most commonly used inotrope was norepinephrine, with a median duration of administration of 1 day (range, 0 to 4 days) in PL patients and zero days (range, 0 to 4 days) in GM patients (p = 0.39). The proportion of patients receiving at least one inotrope at any point during the study infusion period was 6 of 8 (PL) and 5 of 10 (GM) (relative risk of PL versus GM = 1.5, 95% confidence interval, 0.7–3.1, p = 0.37).

ARDS was present on entry in less than half of the patients in both groups. After 5 days, ARDS tended to resolve in those patients treated with GM, but there was a net increase of two ARDS cases with conventional treatment (Table 2).

Detailed prospective measurement of the PaO2/FiO2 ratio at 0, 0.5, 2.5, and 5 hours during each of five daily drug infusions in a subgroup of three GM (1× ARDS on entry) and three PL (2× ARDS on entry) did not detect any early arterial oxygen desaturation associated with GM administration during either the first (p = 0.68) or any subsequent (p = 0.44) infusion (data not shown). In contrast, GM therapy induced repeated, rapid, but transient leukopenia (p = 0.02) that recovered by the end of each 5-hour infusion period (data not shown).

Organ Dysfunction

There was no evidence overall of acute (5 days) or delayed (median ICU duration, 17.5 to 21.5 days) toxicity with GM in the six organ systems (cardiovascular, pulmonary, renal, hepatic, coagulation, and central nervous system) assessed by the sepsis-related organ failure assessment score (Table 2).

Cell Function
Blood leukocyte numbers.

The total number of leukocytes and neutrophils circulating in the peripheral blood of these septic patients was similar on entry (p > 0.2; Table 1). Neutrophil numbers tended to increase with 5 days of GM (mean increase 4.4 × 109/L, p = 0.08) but not PL (mean decrease 0.5 × 109/L, p = 0.35) (Table 2 and Figure 2A)


BAL cell populations.

On trial entry, BAL cell populations (mean ± SD) were similar, being composed of alveolar macrophages (PL 58% ± 28%, GM 43% ± 38%, p = 0.34) and an elevated proportion of neutrophils (PL 29% ± 24%, GM 51% ± 37%, p = 0.23) (Figures 2B and 2C). The alveolar polymorphonuclear leukocyte proportion declined to a mean of 20% ± 17% with PL (p = 0.37) and 22% ± 22% with GM (p = 0.01) by Day 5 (Table 3)

TABLE 3. Statistical summary of the effect of gm-csf on the number and function of blood and alveolar leukocytes

 PL versus GM

 Day 0 versus Day 5

Day × Treatment
Blood neutrophil count0.870.850.09
BAL neutrophil count0.780.010.90
BAL neutrophil proportion0.410.010.13
Blood granulocyte superoxide production0.750.130.02
Blood granulocyte phagocytic function0.960.100.05
Alveolar cell phagocytic function

Definition of abbreviations: BAL = bronchoalveolar; GM = GM-CSF, granulocyte-macrophage colony-stimulating factor; PL = placebo.

Results from significance testing with repeated-measures analysis of variance. Original data are the same as that illustrated in Figures 2 and 3.

. Concurrently, macrophages became the predominant cell type, increasing to 75% ± 19% with PL (p = 0.30) and 70% ± 26% with GM (p = 0.01).

Leukocyte superoxide production and phagocytic function.

The mean superoxide production of stimulated blood granulocytes from PL patients was similar at trial entry compared with 5 days later, but the neutrophils of GM patients had the potential to produce significantly greater superoxide (Figure 3A

and Table 3). Also, peripheral blood granulocytes declined in their ability to phagocytose zymosan following severe sepsis, but GM-CSF therapy tended to prevent this decline (Figure 3B and Table 3). Likewise in the lung, the mixed neutrophils and macrophages recovered by BAL exhibited a tendency to decline in mean phagocytic function in PL that did not occur in GM (Figure 3C and Table 3).

Adverse Events

GM therapy was generally well tolerated, with 9 of 10 courses administered to completion (5 days) without toxicity greater than World Health Organization Grade 1. One 73-year-old subject with sepsis-related acute renal insufficiency (plasma creatinine 0.19 mM at trial entry) experienced transient reversible oliguria without further elevation in plasma creatinine that commenced with the third dose of GM and recurred despite a subsequent 50% dose reduction (World Health Organization Grade 3 renal toxicity). This subject received no dose on Day 5, and the plasma creatinine level returned to baseline (0.12 mM) on discharge from intensive care 1 month later.

This double-blind, PL-controlled phase II trial in 18 patients with severe sepsis and respiratory dysfunction found that the addition of low-dose GM-CSF to conventional intensive medical care was associated with an improvement in oxygenation, which persisted for at least 5 days after the last GM-CSF dose. In addition, there was a closely related improvement in the prevalence of ARDS, with both these effects being of sufficient magnitude to be of potential clinical relevance. The observed improvement in oxygenation was not explained by changes in a number of factors known to influence the physiology of pulmonary gas exchange, including the mechanical ventilation tidal volume and level of continuous positive airway pressure, the alveolar Pco2, or cardiovascular resuscitation. The low-dose GM therapy used in this study was not associated with any extra requirement for inotrope or fluid resuscitation in this group of critically ill patients with sepsis, despite the documented association between GM-CSF therapy, especially at higher doses, and a systemic inflammatory response syndrome (20, 21).

The mechanism by which GM-CSF may induce an improvement in pulmonary gas exchange in patients with severe sepsis and respiratory dysfunction is not able to be determined from this investigation. However, there is increasing evidence that GM-CSF, as a polyfunctional cytokine influencing both neutrophils and monocyte/macrophages, has an important role in pulmonary function beyond that of the granulocyte-lineage–restricted cytokine, G-CSF. In a recent study of patients in the first days of ARDS, elevated amounts of endogenous GM-CSF, but not G-CSF, in BAL fluid were significantly associated with patient survival (22), suggesting a beneficial role for GM-CSF in the pulmonary pathophysiology of ARDS. As well, administration of GM-CSF to patients has been demonstrated to increase the number of circulating granulocytes and monocytes over several days, whereas monocyte activation (23) and monocyte/macrophage phagocytic clearance of apoptotic inflammatory cells, including neutrophils, persists for several weeks (24). In this study, the onset within days of a discernible effect of GM-CSF on pulmonary gas exchange and the persistent gas exchange advantage of GM-CSF–treated patients suggest that part of the mechanism of improvement may involve effects on pulmonary monocyte/macrophages (25), which are myeloid-derived cells (26).

Associated with the improvement in oxygenation following administration of GM-CSF seen in this study was a reduction in the number of pulmonary neutrophils recovered by BAL, whereas peripheral blood neutrophils showed the expected tendency to increase. Of note, a recent study in ARDS patients found that the severity of pulmonary neutrophilia correlated strongly with elevated pulmonary amounts of G-CSF and interleukin-8 but not with the (lower) amounts of GM-CSF that were present (27). This is consistent with other work showing that the precise effects of G-CSF and GM-CSF may result in specifically different modulation of neutrophil inflammatory responses (28, 29). Indeed, in certain circumstances, a combination of G-CSF and GM-CSF may be antagonistic on the functional state of mature cells, such as when GM-CSF negatively modulates G-CSF–initiated increases in neutrophil alkaline phosphatase activity and amino acid incorporation (30).

The syndrome of immunosuppression referred to as the compensatory antiinflammatory response (31) that follows severe sepsis has attracted recent attention as a therapeutic target (2, 3). On entry to this study, many patients had an elevated proportion of neutrophils in their BAL, suggesting the presence of an early phase of acute lung injury (ALI)/ARDS. However, the several-day duration of mechanical ventilation before trial drug commencement also suggests that many patients were likely to have progressed to a mixed systemic inflammatory response syndrome/compensatory antiinflammatory response state at the point when treatment commenced with GM-CSF. In this context, the expected sepsis-associated decline in circulating neutrophil phagocytic and oxidative function associated with the compensatory antiinflammatory response was abrogated or diminished by a low dose of systemically administered GM-CSF. A similar effect was seen on alveolar phagocytes, confirming that the low dose of GM-CSF used in this study was sufficient to exert biologic activity in the lung.

Further elucidation of the mechanism of gas exchange improvement noted in this study will require additional investigations, such as into the effect of GM-CSF on multiple cellular functions (28, 32, 33), including the innate immune clearance of pulmonary (7, 34) or extrapulmonary (35) microbiologic invaders, neutrophil apoptosis at sites of inflammation (29), and upon the production and function of pulmonary surfactant in critical illness. This study adds to an emerging body of evidence that GM-CSF may not be causatively associated with excessive leukocyte activation or tissue influx in sepsis or ARDS (22, 27), by finding an improvement in ARDS and pulmonary function at biologically active doses of GM-CSF. Neutrophil activation related to GM-CSF might still contribute to pulmonary impairment in certain circumstances, as in the first dose effect (20), where high maximum concentrations of GM-CSF are followed by transient pulmonary dysfunction together with a systemic syndrome of vasodilation, hypotension, and capillary leakage of some hours of duration. Of importance in this study, no deterioration in pulmonary function was evident on the first or any subsequent trial day, despite all patients having sepsis-related impairment of gas transfer on entry, slightly under half of whom met consensus criteria for the presence of ARDS (13). Deterioration in pulmonary function may have been avoided in the current trial by the use of both a low-dose and slow-infusion schedule (21). Apart from one patient with sepsis-related renal impairment who received GM-CSF and had reversible oliguria without an increase in serum creatinine, clinically important toxicity from the low-dose GM-CSF therapy was not manifest in the five extrapulmonary organ systems included in the sepsis-related organ failure assessment score.

This study in severe sepsis with respiratory dysfunction demonstrated that GM-CSF administration was generally well tolerated and may lead to an improvement in oxygenation. Furthermore, our results suggest that GM-CSF–related neutrophil activation may not be injurious to the lung in ARDS and that GM-CSF may exert other cellular effects that are beneficial in sepsis-related pulmonary dysfunction. These findings indicate the need for further investigations into the role of GM-CSF in the pathophysiology and clinical management of sepsis-related pulmonary dysfunction and ARDS.

This work was supported by Schering-Plough Pty Ltd., Australia, and the National Health and Medical Research Council, Australia.

1. Waage A. Tumor necrosis factor and septic shock [letter]. Lancet 1998; 351:603.
2. Döcke W-D, Randow F, Syrbe U, Krausch D, Asadullah K, Reinke P, Volk H-D, Kox W. Monocyte deactivation in septic patients: restoration by IFN-γ treatment. Nat Med 1997;3:678–681.
3. Williams MA, White SA, Miller JJ, Toner C, Withington S, Newland AC, Kelsey SM. Granulocyte-macrophage colony-stimulating factor induces activation and restores respiratory burst activity in monocytes from septic patients. J Infect Dis 1998;177:107–115.
4. Flohe S, Borgermann J, Dominguez FE, Majetschak M, Lim L, Kreuzfelder E, Obertacke U, Nast-Kolb D, Schade FU. Influence of granulocyte-macrophage colony-stimulating factor (GM-CSF) on whole blood endotoxin responsiveness following trauma, cardiopulmonary bypass, and severe sepsis. Shock 1999;12:17–24.
5. Lieschke GJ, Burgess AW. Drug therapy: granulocyte colony-stimulating factor and granulocyte-macrophage colony-stimulating factor (first of two parts). N Engl J Med 1992;327:28–35.
6. Stanley E, Lieschke GJ, Grail D, Metcalf D, Hodgson G, Gall JAM, Maher DW, Cebon J, Sinickas V, Dunn AR. Granulocyte-macrophage colony-stimulating factor-deficient mice show no major perturbation of hematopoiesis but develop a characteristic pulmonary pathology. Proc Natl Acad Sci USA 1994;91:5592–5596.
7. LeVine AM, Reed JA, Kurak KE, Cianciolo E, Whitsett JA. GM-CSF-deficient mice are susceptible to pulmonary group B streptococcal infection. J Clin Invest 1999;103:563–569.
8. Seymour JF, Lieschke GJ, Grail D, Quilici C, Hodgson G, Dunn AR. Mice lacking both granulocyte colony-stimulating factor (CSF) and granulocyte-macrophage CSF have impaired reproductive capacity, perturbed neonatal granulopoiesis, lung disease, amyloidosis, and reduced long term survival. Blood 1997;90:3037–3049.
9. Seymour JF, Presneill JJ, Schoch OD, Downie GH, Moore PE, Doyle IR, Vincent JM, Nakata K, Kitamura T, Langton D, et al. Therapeutic efficacy of granulocyte-macrophage colony-stimulating factor in patients with idiopathic acquired alveolar proteinosis. Am J Respir Crit Care Med 2001;163:524–531.
10. Pieters RC, Rojer RA, Saleh AW, Saleh AEC, Duits AJ. Molgramostim to treat SS-sickle cell leg ulcers [letter]. Lancet 1995;345:528.
11. Sweeney JF, Rosemurgy AS, Wei S, Djeu JY. Candida antigen titer is a marker of neutrophil dysfunction after severe injury. J Trauma 1994; 36:797–802.
12. American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992;20:864–874.
13. Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L, Lamy M, LeGall JR, Morris A, Spragg R, et al. The American-European consensus conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994;149:818–824.
14. Moreno R, Vincent JL, Matos R, Mendonca A, Cantraine F, Thijs L, Takala J, Sprung C, Antonelli M, Bruining H, et al. The use of maximum SOFA score to quantify organ dysfunction/failure in intensive care: results of a prospective, multicentre study: working group on sepsis related problems of the ESICM. Intensive Care Med 1999;25:686–696.
15. Wilson JW, Djukanovic R, Howarth PH, Holgate ST. Lymphocytic activation in bronchoalveolar lavage and peripheral blood in atopic asthma. Am Rev Respir Dis 1992;145:958–960.
16. Wahlstrom J, Berlin M, Skold CM, Wigzell H, Eklund A, Grunewald J. Phenotypic analysis of lymphocytes and monocytes/macrophages in peripheral blood and bronchoalveolar lavage fluid from patients with pulmonary sarcoidosis. Thorax 1999;54:339–346.
17. Stewart AG, Harris T, De Nichilo M, Lopez AF. Involvement of leukotriene B4 and platelet-activating factor in cytokine priming of human polymorphonuclear leukocytes. Immunology 1991;72:206–212.
18. Pereira HA, Shelton MJ, Hosking CS. Neutrophil iodination micro-method as an index of neutrophil and opsonic function. J Clin Lab Immunol 1983;11:47–53.
19. Armitage P, Berry G. Statistical methods in medical research, 3 ed. Oxford: Blackwell Scientific; 1994. p. 386–394.
20. Lieschke GJ, Cebon J, Morstyn G. Characterization of the clinical effects after the first dose of bacterially synthesized recombinant human granulocyte-macrophage colony-stimulating factor. Blood 1989;74:2634–2643.
21. Cebon J, Lieschke GJ, Bury RW, Morstyn G. The dissociation of GM-CSF efficacy from toxicity according to route of administration: a pharmacodynamic study. Br J Haematol 1992;80:144–150.
22. Matute-Bello G, Liles WC, Radella F, Steinberg KP, Ruzinski JT, Hudson LD, Martin TR. Modulation of neutrophil apoptosis by granulocyte colony-stimulating factor and granulocyte/macrophage colony-stimulating factor during the course of acute respiratory distress syndrome. Crit Care Med 2000;28:1–7.
23. Williams MA, Kelsey SM, Collins PW, Gutteridge CN, Newland AC. Administration of rHuGM-CSF activates monocyte reactive oxygen species secretion and adhesion molecule expression in vivo in patients following high dose chemotherapy. Br J Haematol 1995;90:31–40.
24. Galati G, Rovere P, Citterio G, Bondanza A, Scagliette U, Bucci E, Heltai S, Fascio U, Rugarli C, Manfredi AA. In vivo administration of GM-CSF promotes the clearance of apoptotic cells: effects on monocytes and polymorphonuclear leukocytes. J Leukoc Biol 2000;67:174–182.
25. Rosseau S, Hammerl P, Maus U, Walmrath HD, Schutte H, Grimminger F, Seeger W, Lohmeyer J. Phenotypic characterization of alveolar monocyte recruitment in acute respiratory distress syndrome. Am J Physiol Lung Cell Mol Physiol 2000;279:L25–L35.
26. Thomas ED, Ramberg RE, Sale SE. Direct evidence for a bone marrow origin of the alveolar macrophages of man. Science 1976;192:1016–1018.
27. Aggarwal A, Baker CS, Evans TW, Haslam PL. G-CSF and IL-8 but not GM-CSF correlate with severity of pulmonary neutrophilia in acute respiratory distress syndrome. Eur Respir J 2000;15:895–901.
28. Pitrak DL. Effects of granulocyte colony-stimulating factor and granulocyte-macrophage colony-stimulating factor on the bactericidal functions of neutrophils. Curr Opin Hematol 1997;4:183–190.
29. Hu B, Yasui K. Effects of colony-stimulating factors (CSFs) on neutrophil apoptosis: possible roles at inflammation site. Int J Hematol 1997; 66:179–188.
30. Teshima T, Shibuya T, Harada M, Akashi K, Taniguchi S, Okamura T, Niho Y. Granulocyte-macrophage colony-stimulating factor suppresses induction of neutrophil alkaline phosphatase synthesis by granulocyte colony-stimulating factor. Exp Hematol 1990;18:316–321.
31. Bone RC. Sir Isaac Newton, sepsis, SIRS, and CARS. Crit Care Med 1996;24:1125–1128.
32. Shibata Y, Berclaz PY, Chroneos ZC, Yoshida M, Whitsett JA, Trapnell BC. GM-CSF regulates alveolar macrophage differentiation and innate immunity in the lung through PU.1. Immunity 2001;15:557–567.
33. Trapnell BC, Whitsett JA. GM-CSF regulates pulmonary surfactant homeostasis and alveolar macrophage-mediated innate host defense. Annu Rev Physiol 2002;64:775–802.
34. Paine R III, Preston AM, Wilcoxen S, Jin H, Siu BB, Morris SB, Reed JA, Ross G, Whitsett JA, Beck JM. Granulocyte-macrophage colony-stimulating factor in the innate immune response to Pneumocystis carinii pneumonia in mice. J Immunol 2000;164:2602–2609.
35. Garcia-Gonzalez M, Boixeda D, Herrero D, Burgaleta C. Effect of granulocyte-macrophage colony-stimulating factor on leukocyte function in cirrhosis. Gastroenterology 1993;105:527–531.
Correspondence and requests for reprints should be addressed to Dr. Jeffrey J. Presneill, Intensive Care Unit, The Royal Melbourne Hospital, Victoria 3050, Australia. E-mail:


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record