IN THIS ISSUE of the Archives, The National Committee for the Evaluation of Centoxin1 reports data from a French national registry that monitored the outcome of 600 septic patients given a single 100-mg dose of HA-1A (Centoxin, Centocor, Malvern, Pa), a sepsis drug that was recently approved for, and then withdrawn from, clinical use in Europe. The authors show that patients treated with HA-1A had a higher mortality rate than would have been expected from their scores measured by means of the Acute Physiology and Chronic Health Evaluation II (APACHE II). The study was retrospective and compared HA-1A\p=m-\associatedmortality with APACHE II\p=m-\predictedmortality, rather than a simultaneously treated control group. Although the P value of .06 for increased mortality did not reach statistical significance, P values less than . 1 are more than adequate to suggest evidence of harm. Despite these potential limitations, the French findings should prompt a reexamination of the sequence of events leading to the rapid ap¬ proval and short-lived clinical use of HA-1A for sepsis.Nearly 10 years ago, investigators in the United States2 reported that HA-1A bound specifically to lipid A, the toxic part of endotoxin that is structurally similar among all gramnegative bacteria. These investigators also reported that See also page 2484 HA-1A bound to endotoxin obtained from multiple spe¬ cies of gram-negative bacteria and that it improved sur¬ vival in a mouse model of sepsis. ses were available to staff members at Centocor Inc (the manufacturer of HA-1A and sponsor of the clinical trial) before completion of the trial. After these interim results were available to Centocor, end points in the original ana¬ lytic plan were modified. If the data had been analyzed ac¬ cording to the original plan, HA-1A would not have pro¬ duced a significant effect on survival.Questions were also raised about preclinical data. In¬ dependent investigators in Europe found that HA-1A bind¬ ing properties were not specific for endotoxin and that HA-1A binds to a broad variety of antigens.6 The re-ported protective effect of HA-1A in small animal models of sepsis could not be confirmed.7 Furthermore, a blinded, controlled trial of HA-1A in dogs with gram-negative sep¬ tic shock showed significant harmful effects.8 Because of the above controversies,3-9 a second multicenter, random¬ ized, double-blind,