Résumé : OBJECTIVE:: This trial compared the efficacy/safety of two IV doses of AZD9773, a polyclonal antibody to tumor necrosis factor-α, in adult patients with severe sepsis/septic shock. DESIGN:: Multicenter, randomized, double-blind, placebo-controlled phase IIb trial. SETTING:: ICUs in seven countries (Australia, Belgium, Canada, Czech Republic, Finland, France, and Spain). PATIENTS:: Patients 18 years old or older with severe sepsis and/or septic shock. Patients were required to have 1) objective clinical evidence of infection; 2) at least two of four systemic inflammatory response syndrome criteria; and 3) cardiovascular and/or respiratory sepsis-related failure. INTERVENTIONS:: Patients were randomized 1:1:1 to a single loading infusion of AZD9773 250 U/kg followed by 50 U/kg every 12 hours (low dose, n = 100), a single loading infusion of AZD9773 500 U/kg followed by 100 U/kg every 12 hours (high dose, n = 100), or placebo (n = 100) for 5 days. Follow-up assessments were performed up to day 90. MEASUREMENTS AND MAIN RESULTS:: Mean number of ventilator-free days (primary endpoint) did not differ between low-dose (19.7 d) or high-dose AZD9773 (17.3 d) and placebo (18.3 d) (one-sided p = 0.18 and 0.74, respectively). Mortality rates were comparable across treatment groups; relative risk of death versus placebo at day 29 was 0.80 for low-dose AZD9773 (one-sided p = 0.25) and 1.64 for high-dose AZD9773 (p = 0.97). Most patients experienced at least one treatment-emergent adverse event (87.8% in AZD9773-treated patients, 92.9% in placebo patients) although most were mild/moderate in nature. No differences in the incidence of adverse events or laboratory or vital sign abnormalities were observed between groups. CONCLUSIONS:: AZD9773 rapidly and efficiently decreased plasma tumor necrosis factor-α concentration in patients with severe sepsis/septic shock, but this effect did not translate into clinical benefit. Copyright © 2013 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins.