The effects of granulocyte colony-stimulating factor in preclinical models of infection and acute inflammation.

The cytokine granulocyte colony-stimulating factor (G-CSF) is a potent endogenous trigger for the release of neutrophils from bone marrow stores and for their activation for enhanced antimicrobial activity. G-CSF has been widely evaluated in preclinical models of acute illness, with generally promising though divergent results. A recombinant G-CSF molecule has recently undergone clinical trials to assess its efficacy as an adjuvant therapy in community-acquired and nosocomial pneumonia, however, these studies failed to provide convincing evidence of benefit. We undertook a systematic review of the published literature reporting the effects of modulation of G-CSF in preclinical in vivo models to determine whether evidence of differential efficacy might explain the disappointing results of human studies and point to disease states that might be more likely to benefit from G-CSF therapy. G-CSF has been evaluated in 86 such studies involving a variety of different models. The strongest evidence of benefit was seen in studies involving intraperitoneal challenge with live organisms; benefit was evident whether the agent was given before or after challenge. G-CSF demonstrates anti-inflammatory activity in models of systemic challenge with viable organisms or endotoxin, but only when the agent is given before challenge; evidence of benefit after challenge was minimal. Preclinical models of intrapulmonary challenge only show efficacy when the cytokine is administered before the infectious challenge, and suggested harm in gram-negative pneumonia resulting from challenge with Escherichia coli or Klebsiella. There is little evidence for therapeutic efficacy in noninfectious models of acute illness. We conclude that the most promising populations for evaluation of G-CSF are neutropenic patients with invasive infection and patients with intra-abdominal infection, particularly those with the syndrome of tertiary, or recurrent, peritonitis. Significant variability in the design and reporting of studies of preclinical models of acute illness precludes more sophisticated data synthesis.