Sunday, September 23, 2007

Activated protein C in severe sepsis: patient selection is getting harder

A meta-analysis recently published in BioMed Central Anesthesiology raised troubling questions on the use of recombinant human activated protein C (rhAPC) in severe sepsis.

First a little background. Sepsis is defined by known or suspected infection in the presence of the systemic inflammatory response syndrome (SIRS). Sepsis is defined as being severe when accompanied by dysfunction of at least one organ system. For decision making purposes clinicians attempt to further subdivide cases of severe sepsis into those with a lower risk of death and those with the highest risk of death.

Since rhAPC was approved in 2001 it has been the subject of controversy because its approval was based on a subset analysis of a single randomized controlled trial, PROWESS. The patients in PROWESS determined to have the highest risk of death based on an APACHE-II score of at least 25 or involvement of 2 or more organs accounted for the reduction in mortality associated with rhAPC.

The FDA required a follow up study of rhAPC in patients with severe sepsis defined as having a lower risk of death. This study, ADDRESS, was halted early due to futility. On the basis of PROWESS and ADDRESS rhAPC was recommended only for patients with severe sepsis who had a high risk of death.

ENHANCE was a single arm open label post marketing study to evaluate the efficacy and safety of rhAPC in conditions resembling real world clinical practice, using patient selection criteria similar to PROWESS. No distinction was made among patients with severe sepsis between those having a high or low risk of death although the vast majority of patients had dysfunction of 2 or more organs. The mortality in ENHANCE was almost identical to that observed in the PROWESS treatment arm although the rate of bleeding complications was higher. A notable finding in ENHANCE was that patients treated within the first 24 hours of the first evidence of organ dysfunction had a lower mortality than those treated later.

The BioMed Central Anesthesia meta-analysis raised concerns about the efficacy of rhAPC and proper patient selection. The authors performed a pooled analysis of PROWESS and ADDRESS to assess the effect on mortality, and included several other studies in a cost effectiveness analysis. There was no statistically significant overall effect on mortality (RR .93, CI .69-1.26). That finding represents the overall patient population and is not particularly surprising considering that the less severely ill ADDRESS patients, for whom rhAPC is off label, were included.

But that’s not all. The subgroup analysis produced worrisome findings. In order to understand the subgroup analysis it is important to emphasize that the ADDRESS cohort of less ill patients contained some with APACHE-II scores over 25 and some with multiple organ dysfunction. That confusing situation arose because labeling criteria for defining patients at high risk of death, which were used to select patients for ADDRESS, tended to be vague and varied among participating countries.

Here’s the graphic representation of the subgroup analysis for the pooled results of ADDRESS and PROWESS. The findings of interest were these:

APACHE-II score of 25 or greater: RR .9, CI .54-1.49.
Two or more dysfunctional organs: RR .84, CI .70-1.00

These results may be skewed by the patients in address with APACHE-II scores of 25 or greater, in which the RR was 1.19 (CI .83-1.71).

How can the different results in the ADDRESS patients with multiple organ dysfunction or high APACHE-II scores be explained? First, these were probably less severely ill patients than those in PROWESS. Secondly, ADDRESS study sites included many smaller community hospitals where the level of investigator expertise may have been low. Complex and potentially dangerous drugs like rhAPC generally perform better in the hands of expert clinical trialists than they do in the community.

In light of this new information, what is the status of rhAPC (Xigris) today? My impressions are as follows:

1) Xigris can ethically be considered for additional clinical trials.
2) Xigris is a niche drug for more gravely ill patients with severe sepsis. A simple formulaic approach to patient selection based on the APACHE-II score is not sufficient. The APACHE-II score must be coupled with clinical judgment. Suitable candidates might include patients with both a high APACHE-II score and multiple organ dysfunction.
3) Clinical expertise is important for optimal use of Xigris. Hospitals and health care systems may wish to restrict its use to critical care specialists.
4) Early use of Xigris is important for optimal results. Benefits diminish beyond 24 hours after the first evidence of organ dysfunction.
5) Although Xigris has a role in patients with severe sepsis and can be a life saving drug when used in the optimal patient population and with appropriate expertise it should not, in my opinion, be considered standard of care.
6) New information about rhAPC presents a difficult challenge to the writers of the next version of the Surviving Sepsis guidelines.

This paper gives me yet another reason to appreciate BioMed Central: it’s apparently off the radar screen for popular media outlets such as the New York Times. Thankfully, nobody there seems to have noticed the paper, which could have been published in NEJM or JAMA. If it had, the media would have pounced on it. I can just see the headline: “Eli Lilly’s expensive sepsis drug ineffective”.

No comments: