Yesterday was a great day for evidence-based medicine. In light of what should be the definitive nail in the coffin, Eli Lilly has decided to pull Xigris (drotrecogin alfa activated) from the market. This drug represented an innovative and promising approach to the treatment of severe sepsis, but it seemed doomed from the outset. I'm certain that I am not alone in my amazement that it has been on the market for so long.
Sepsis is an overwhelming systemic inflammatory response to infection, characterized by high systemic levels of proinflammatory cytokines, such as IL-1b, IL-6, IL-8, and TNF-a. By numerous mechanisms - some known, probably more unkown, this results in a variety of organ system failures - cardiovascular, respiratory, hematologic, neurologic, renal, and hepatic. The mortality rate ranges, depending on the number and severity of organ failures, from 20 to 60 percent. Treatment of sepsis is largely comprised of treatment of the underlying infection and supportive care (fluids, vasopressors, artificial respiration, dialysis) for organ failures. Naturally, when a novel therapy for severe sepsis came along, the critical care community, desperate for a new idea, embraced it with open arms.
The theory behind drotrecogin, which is a recombinant human activated protein C (rhAPC), is that by modifying the complex interaction between the clotting cascade, the intrinsic fibrinolytic system, and the inflammatory response, outcomes in sepsis could be improved. Activated protein C inhibits coagulation by inactivating factors V and VII leading to reduction in thrombin generation. It promotes fibrinolysis by inactvating PAI-1 (Plasminogen Activation Inhibitor) and preventing activation of TAFI (Thrombin Activatiable Fibrinolysis Inhibitor). Finally, it exerts antiinflammatory effects by supression of adhesion (also dependent upon the clotting cascade) and cytokine release by activated leukocytes.
PROWESS was an industry-funded (Eli Lilly), multicenter phase III clinical trial in which patients with severe sepsis (sepsis with at least one organ failure) were randomized to receive a 96 hour infusion of rhAPC or placebo. The results were impressive - a reduction in mortality from 30.8% in the placebo group to 24.7% in the treatment group. Almost immediately, however, the story began to unravel.
PROWESS is likely the most widely criticized clinical trial in the critical care literature. The relatively small size of the trial is of concern. More importantly, however, the study protocol was changed midstream. They changed the master cell bank from which the recombinant protein was derived. The eligibility criteria were amended to discourage enrollment of patients with serious pre-existing illness. Even the placebo was changed. Curiously, the efficacy of treatment with rhAPC was greater after these changes were made. Despite the above concerns, and despite the split vote by the FDA advisory panel (10-10), the FDA elected to approve rhAPC for use in the treatement of severe sepsis.
The more deeply this drug was studied, the less support it found. Post-hoc analyses suggested that less ill patients may actually have an increased risk of harm, be it serious bleeding or even death. The ADDRESS trial prospectively studied rhAPC in patients with low risk of death. Not only were the mortality rates similar betwen groups (17% vs. 18.5%), but mortality in the placebo group was substantially lower than that of similar placebo patients in the PROWESS trial. ENHANCE was an open label trial to look in more detail about the mortality and morbidity rates in treated patients; mortality was similar to PROWESS, but the risk of severe bleeding events was significantly increased. A subsequent large meta-analysis pooled data from five major trials and found no statistically significant benefit to the administration of rhAPC in severe sepsis.
In 2007, the European Medicines Agency (EMA) determined that further clarification of the risk/benefit balance of rhAPC was required. The PROWESS-SHOCK trial addressed this issue. It enrolled similar patients to those in the original PROWESS study, but with more stringent criteria. Patients were required to be in vasopressor-dependent septic shock through the 8-hour pretreatment and subsequent enrollment periods, and were required to have at least one other manifestation of end-organ hypoperfusion (decreased urine output, metabolic acidosis, hepatic dysfunction). Interim analyses were permitted for adjustment of sample size, but early efficacy termination was not permitted after the first analysis (as it was in PROWESS). Futility termination was also not permitted, but termination due to apparent harm was at the discretion of the data monitoring committee.
This study is, perhaps, the most rigorously designed trial in the history of critical care. The manuscript has yet to be published, but the results are incontrovertible - 26.4% mortality in the rhAPC vs. 24.2% in the placebo group. This validates the concerns of critics of PROWESS and confirms the findings of numerous trials and meta-analyses published since PROWESS. As a result, Lilly has removed Xigris from the market. Lilly's Chief Medical Officer, Dr. Timothy Garnett understandably, and incorrectly, identifies improvement in the standard of care for patients with severe sepsis as the reason for the disparity between PROWESS and subsequent trials. Were that the case, the improved standard of care would apply to both groups, lowering overall mortality while preserving the statistically significant difference between them. If anything, the opposite hapened - a statistically insignificant trend towards harm associated with the administration of rhAPC.
I sincerely hope that the FDA stands up and takes notice - and responsibility for their mistake. This is only the latest in a long series of black marks against an agency that seems to be able to do no right. Clearly, the drug approval process needs to be modified, and if nothing else, approval needs to be conditional pending confirmatory studies and intensive post-approval surveillance.