A meta-analysis of controlled trials of recombinant human activated protein C therapy in patients with sepsis

BACKGROUND: Meta-analysis of two randomised controlled trials in severe sepsis performed with recombinant human activated protein C may provide further insight as to the therapeutic utility of targeting the clotting cascade in this syndrome. METHODS: In search for relevant studies published, two randomized clinical trials were found eligible. RESULTS: The studies, PROWESS and ADDRESS, enrolled a total of 4329 patients with risk ratio (RR) and 95% confidence interval (CI) data for effect on 28-day mortality relative to control treatment of 0.92 (0.83–1.02) suggesting that recombinant human activated protein C is not beneficial in severe sepsis. In PROWESS, 873 of 1690 patients presented with low risk, and 2315 of 2639 patients in ADDRESS as defined by APACHE II score < 25. In this low-risk stratum, no effect of recombinant human activated protein C administration on 28-day mortality was observed. This observation appears to be consistent and homogenous. Heterogeneity between the two studies, however, was seen in patients with APACHE II score ≥ 25 in whom recombinant activated protein C was effective in PROWESS (n = 817; RR 0.71, CI 0.59–0.85) whereas a tendency toward harm was present in ADDRESS (n = 324; RR 1.21, CI 0.85–1.74). Even though the overall treatment effect in this high-risk population was still in favour of treatment with recombinant activated protein C (n = 1141; RR 0.80, CI 0.68–0.94), the observed heterogeneity suggests that the efficacy of recombinant human activated protein C is not robust. Not unlikely, the adverse tendency observed could have become significant with higher statistical power would ADDRESS not have been terminated prematurely. CONCLUSION: This meta-analysis, therefore, raises doubts about the clinical usefulness of recombinant activated protein C in patients with severe sepsis and an APACHE II score ≥ 25 which can only be resolved by another properly designed clinical trial.