Anti-inflammatory therapies to treat sepsis and septic shock: A reassessment
This issue of Critical Care Medicine reports the results of the second phase III double-blinded multicenter trial of interleukin-1 receptor antagonist (IL-1ra) . In a clear presentation and discussion of the data, Dr. Opal and colleagues  conclude that the drug had no significant clinical effect in septic patients. The trial was stopped early because an interim analysis determined that the drug was unlikely to show benefit if the study were completed. Before the trial was terminated, 906 septic patients were enrolled. Four hundred fifty patients received IL-1ra, with a 34% mortality rate compared with 36% in controls. In the first IL-1ra phase III double-blinded trial, 893 patients were enrolled and 591 patients received IL-1ra, with a 30% mortality rate compared with 34% in controls . The remarkably consistent results of these two large studies suggest that if IL-1ra has a beneficial effect, it is small, probably less than a 10% decrease in all-cause mortality rate during sepsis and septic shock.
A number of other nonglucocorticoid anti-inflammatory agents have also failed to show significant benefit in septic patients . Placing the IL-1ra results in this broader context might provide insight about the future of developing therapeutic strategies based on inhibiting the inflammatory response in septic patients. In particular, this larger database could help us understand why IL-1ra and other anti-inflammatory agents failed, and whether this approach to sepsis is beneficial, harmful, or has no effect. Toward this end, a computerized search of the medical literature from 1986 to the present (using Embase and Medline) and a search of the proceedings of recent industry and scientific meetings were undertaken in order to identify studies that, like these two IL-1ra trials [1,2], used nonglucocorticoid agents with a purported anti-inflammatory mechanism to treat sepsis.
Data from any clinical trial found in this search that included a group that could be classified as control were then introduced into the meta-analysis. For studies involving multiple doses of a drug without a control agent, the patients receiving the lowest doses were designated as the controls. To simplify the analysis, doses were pooled whenever treatment effects were similar. If significant differences were found among treatment effects of various doses, the results for individual doses were compared with all other studies analyzed and, if similar, pooled. In all, 19 clinical trials were identified. Including the two IL-1ra trials cited above [1,2], there are a total of 21 clinical trials of nonglucocorticoid anti-inflammatory agents in patients with sepsis and septic shock. Of these trials, three phase I studies were excluded because they lacked either survival data or a control group [4-6]. In the 18 remaining clinical trials, septic patients were treated with six different agents: a) bradykinin antagonists [7,8]; b) platelet-activating factor antagonists [9,10]; c) monoclonal antibodies against tumor necrosis factor (TNF) [11-16]; d) soluble TNF receptors [17,18]; e) prostaglandin antagonists [19-21]; and f) IL-1ra [1,2,22] (Table 1).
While most of these clinical trials were double-blinded, five trials were open-label [11-14,22]. A total of 11 studies explored the effects of altering the dose [2,7,11-18,22]. In nine of these studies, the effects of the dose were similar and were averaged. Two doses of anti-TNF monoclonal antibody in one clinical trial had effects on survival that were significantly different from each other . However, these effects were not significantly different from the effects of other agents in the meta-analysis, and were therefore averaged.