Randomized, placebo-controlled trial of the anti-tumor necrosis factor antibody fragment afelimomab in hyperinflammatory response during severe sepsis: The RAMSES Study

Objective: This study investigated whether treatment with the anti-tumor ne crosis factor-or monoclonal antibody afelimomab would improve survival in s eptic patients with serum interleukin (IL)-6 concentrations of >1000 pg/ml, Design: Multicenter, double-blind, randomized, placebo-controlled study. Setting: Eighty-four intensive care units in academic medical centers in Eu rope and Israel. Patients: A total of 944 septic patients were screened and stratified by the results of a rapid qualitative immunostrip test for seru m IL-6 concentrations. Patients with a positive test kit result indicating IL-6 concentrations of >1000 pg/mL were randomized to receive either afelimomab (n = 224) or place bo (n = 222), Patients with a negative IL-6 test (n = 498) were not randomi zed and were followed up for 28 days. Interventions: Treatment consisted of 15-min infusions of 1 mg/kg afelimoma b or matching placebo every 8 hrs for 3 days. Standard surgical and intensi ve care therapy was otherwise delivered. Measurements and Main Results:The study was terminated prematurely after an interim analysis estimated that the primary efficacy end points would not be met. The 28-day mortality rate in the nonrandomized patients (39.6%, 197 of 498) was significantly lower (p < .001) than that found in the randomiz ed patients (55.8%, 249 of 446). The mortality rates in the IL-6 test kit p ositive patients randomized to afelimomab and placebo were similar, 54.0% ( 121 of 224) vs, 57.7% (128 of 222), respectively. Treatment with afelimomab was not associated with any particular adverse events. Conclusions: The IL-6 immunostrip test identified two distinct sepsis popul ations with significantly different mortality rates. A small (3.7%) absolut e reduction in mortality rate was found in the afelimomab-treated patients. The treatment difference did not reach statistical significance.