Tuesday, October 04, 2005

Xigris in perspective

The ADDRESS trial of activated protein C (Xigris) was just published in NEJM. Though it was touted in the press as a breaking development it was really nothing new. Maybe September 28 was a slow news day in healthcare. I wouldn’t have bothered posting it here, but now that it’s been hyped some perspective is needed.

ADDRESS is old hat not only because it was announced months ago but also because it merely confirms what we already knew from a subset analysis of the earlier PROWESS trial: that the optimal use of Xigris is exactly according to the way FDA has labeled it and the company (Eli Lilly) has promoted it all along. Specifically, it is indicated as an adjunct to antibiotics and general supportive care in patients with severe sepsis who have an APACHE-2 score of at least 25.

FDA approval of Xigris was based on PROWESS, analysis of which revealed that the benefits of Xigris were confined to the subset of patients with the higher APACHE-2 scores. The FDA restricted its approval to this group of patients and then required Lilly to conduct a randomized controlled trial (ADDRESS) to specifically assess those patients with lower scores. As anticipated from PROWESS, ADDRESS found no overall survival difference. Although there was no overall difference a trend toward increased mortality with Xigris was seen in patients with recent surgery, as had previously been gleaned from PROWESS in the lower risk patients. Increased mortality was also observed with Xigris in the subset of patients who were the first enrollees at their respective research sites, suggesting a learning curve with the drug, and highlighting the importance of experience. This finding bolsters the suggestion of some that the use of Xigris be restricted to intensivists or other designated specialists.

Unfortunately the study was somewhat hyped in the press. Xigris looked like a flop in some reports and like poison in others, not to mention that most of the news reports were just plain confusing.

No comments: