Granulocyte colony-stimulating factor versus placebo in addition to penicillin G in a randomized blinded study of gram-negative pneumonia sepsis: analysis of survival and multisystem organ failure

Sepsis is a common cause of morbidity and mortality. Neutrophils are the major defense against bacterial invasion, and granulocyte colony- stimulating factor (G-CSF) augments both neutrophil number and function. In our study, 160 rabbits were inoculated transtracheally with 0.5 mL of a solution containing 10(4) colony forming units per milliliter of Pasteurella multocida. Twenty-four hours later, chest x- rays and quantitative blood cultures demonstrated pneumonia and bacteremia. Therapy was then begun with penicillin G and either recombinant human G-CSF (rG-CSF; 5 to 8 micrograms/kg subcutaneously) or placebo every day for 5 days. Arterial blood gases and 23 other parameters of organ function were performed before inoculation and serially thereafter. All rabbits underwent histologic examination of organs at the time of septic death or when sacrificed on day 6. A total of 149 rabbits survived long enough to initiate therapy. A significant increase in leukocytes by day 4 was found in the rG-CSF-treated group. There was a trend towards improved survival in the rG-CSF group (77% v 67%; P = .13, n = 149). Analysis of pretreatment variables revealed sepsis-induced leukopenia (< or = 2,800/microL) as the only predictor of significantly improved survival with rG-CSF treatment (57% v 39%; P = .04, n = 73). The majority of the survival benefit occurred within the first 24 hours of treatment. This was before the time that a significant difference in mean white blood cell (WBC) count was observed between the study groups, making intravascular leukocytosis an unlikely explanation for the survival advantage in the rG-CSF group. No significant difference in laboratory variables reflecting organ function was demonstrated between the groups. Histologic grading of inflammation (0, normal, to 6, necrosis) in seven organs revealed that the surviving rabbits had mild but statistically significant increased inflammation in the liver, spleen, and noninoculated lung in the rG-CSF versus placebo groups (liver: 2.6 v 1.5, P < or = .0001; spleen: 3.2 v 2.3, P < or = .0001; and noninoculated lung: 2.9 v 2.5, P = .04). Administration of rG-CSF, in addition to penicillin G, in immune competent rabbits with gram-negative sepsis complicated by leukopenia significantly improved survival over antibiotics alone. The administration of rG-CSF in early sepsis for a short therapeutic duration was not associated with any clinically evident toxicity. Clinical trials using rG-CSF in septic patients with leukopenia are indicated.