Endotoxin is a primary trigger of the inflammatory processes that lead to shock, multiorgan failure, and purpura fulminans in meningococcal sepsis. Bactericidal/permeability-increasing protein (BPI) is a natural protein, stored within the neutrophil granules, that binds to and neutralises the effects of endotoxin in vitro, in laboratory animals, and in humans. To establish whether a recombinant 21-kDa modified fragment of human BPI (rBPI₂₁), containing the active antimicrobial and endotoxin-neutralising moiety, would decrease death and long-term disability from meningococcal sepsis, we did a randomised, double-blind, placebo-controlled trial of rBPI21 in children with severe meningococcal sepsis.

We enrolled children (2 weeks to 18 years of age) presenting to 22 centres in the UK and the USA with a clinical picture suggestive of meningococcal sepsis, and with evidence of severe disease. Children were randomly assigned rBPI₂₁ (2 mg/kg over 30 min followed by 2 mg/kg over 24 h) or placebo (0·2 mg/mL human albumin solution) in addition to conventional medical therapy. Primary outcome variables were mortality, amputations, and change in paediatric overall performance category (POPC) from before illness to day 60. Analysis was by intention to treat.

Of 1287 patients screened, 892 were excluded, including 57 patients who died or who met criteria for imminent death before receiving the study drug. 190 patients received rBPI21, and 203 placebo. 34 (8·7%) of 393 patients died during the study: 14 (7·4%) in the rBPI21 group and 20 (9·9%) in the placebo group (odds ratio 1·31 [95% Cl 0·62–2·74], p=0·48). Compared with patients randomised to placebo, fewer patients treated with rBPI21 had multiple severe amputations (six of 190 [3·2%] vs 15 of 203 [7·4%], odds ratio 2·47 [0·94–6·51], p=0·067), and more had a functional outcome similar to that before illness (as measured by the POPC scale) at day 60 (136 of 176 [77·3%] vs 126 of 190 [66·3%], p=0·019).

Because most deaths occurred in the interval between identification of patients and study drug administration, the mortality rate in the placebo group was substantially lower than predicted. The trial was therefore underpowered to detect significant differences in mortality. However, patients receiving rBPI21 had a trend towards improved outcome in all primary outcome variables. Given the excellent severity match between placebo and rBPI21 groups at study entry, the results overall indicate that rBPI21 is beneficial in decreasing complications of meningococcal disease.