Severe sepsis: Is activated protein C always the answer?

November 2002

During the last several months, the news media have been full of reports about the rapidly rising cost of health care and, more to the point, the dramatic escalation of the cost of health care premiums being charged by insurers. While employers are picking up part of the tab, employees themselves are contemplating increases of up to 40%-50% in the cost of their health care premiums, or a substantial reduction in benefits, increases in co-pays, and the like.

Against that backdrop, four articles in the Sept. 26, 2002, issue of The New England Journal of Medicine are of more than usual interest. They all deal with attempting to assess the proper role for recombinant activated protein C in the management of severe sepsis and include a cost-benefit analysis, two “Sounding Board” commentaries and a brief perspective authored by Dick Wenzel. For better or for worse, the use of activated protein C in severe sepsis is usually not a decision made by infectious disease physicians, save for the intrepid few who are pulmonary disease physicians or intensivists as well; rather, it is usually made in intensive care units by attending physicians and/or intensivists shortly after admission.

Recall that the key study, Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) trial was published in the NEJM by Bernard et al in March 2002 and demonstrated an almost 20% reduction in the risk of death by day 28, the primary end-point. Recall also that the product was approved for use by the FDA in November 2002 and marketed with the generic name of drotrecogin alfa (Xigris, Eli Lilly). Finally, recall that the product is not inexpensive, costing $6,800 per therapeutic course.

The economic analysis reported in the Sept. 26 NEJM was carried out by Manns et al, at the University of Alberta and the University of Calgary, in Edmonton and Calgary, respectively, in Canada. This was a modeling study, based on data derived from their own data on 819 patients admitted to one of three tertiary care facilities in Calgary during a three-year period from 1996-1999. Using that clinical database, they then modeled an estimated effect of using activated protein C in management of these patients using two differing sets of safety and efficacy assumptions. The first set of assumptions was derived from the published trial of Bernard et al; the second set of assumptions was based on a post hoc re-analysis of data derived from that study that was carried out by the FDA.

Some interesting differences emerged from these two assumptions. In the PROWESS trial, the results seemed to show efficacy across all patient subclasses; the cost per year of life gained by treating all patients was $27,936, though it was modestly more cost-effective to treat patients with an APACHE II score of >25 ($24,484) than those with lower APACHE II scores ($35,632).

The FDA re-analysis, however, indicated that efficacy was essentially limited to patients with an APACHE II score of < 25; the cost per year of life gained in that group using the FDA re-analysis was $19,723; in patients with lower APACHE II scores the cost per year of life gained was a stunning $575,054! The authors further pointed out that the per-year cost of life gained increased sharply as life expectancy decreased, most usually as a result of underlying disease. Thus, they conclude that the drotrecogin is indeed cost-effective in patients with severe sepsis, greater severity of illness and a “reasonable” life expectancy, and they point out that serious consideration needs to be given to broader use of the product in settings in which it is less cost-effective.

The situation gets more complicated, however, when reading the two commentaries. The first, by Warren and co-authors from Massachusetts General Hospital, the NIH and the University of Texas Southwestern Medical Center, points out that there were certain changes made in the study protocol and in the product preparation that appeared to divide the study into two distinct phases in which the efficacy results were somewhat different. Regarding the study protocol, changes were made in the entry criteria that appeared to shift the composition of the study population to patients with less severe underlying illness. Thus, bone marrow or solid organ transplant recipients were now excluded, as were all patients with metastatic cancer, pancreatitis or patients in whom “a commitment for aggressive therapy” was not obtained.

The second change was the introduction of a new master lot of cells in which drotrecogin is produced. Co-incident with these changes was an apparent increase in the observed efficacy of the product. The authors further pointed out that the APACHE II score is not as reproducibly quantitative a determination as one might wish, that inter- and even intra-observer variation is quite common, and that there are legitimate questions about when the APACHE II score, a dynamic variable, ought best be determined. After pointing out that the FDA Anti-infective Drugs Advisory Committee, in considering this product approval last October, actually split 10–10 as to whether it considered drotecogin to be safe and effective, it concluded by arguing that use of the product in severe sepsis should not be considered a “standard of care” until further studies in various subgroups were carried out.

The second commentary, by Jay Siegel of the FDA, basically defends the FDA decision to license the product. He described two analyses to support his argument that the protocol amendments did not result in improved efficacy, but rather tended to exclude patients that appeared more likely to benefit from therapy; thus serious pre-existing disease should not be a reason to withhold drotrecogin therapy.

He described further the multiple exhaustive comparisons of the product made in cells derived from both the old and the new cell bank carried out by both the FDA and Eli Lilly; none were found. He concluded that it was unlikely that this change accounted for any change in efficacy. In his view, the only acceptable time to evaluate the APACHE II score was the same way it was done in the PROWESS trial; that is, the worst observed values in the 24 hours prior to starting drotecogin, and independent of any changes in score brought about by the resuscitative efforts.

It is obvious that the last words about when to use activated protein C have yet to be written. Dick Wenzel, in his perspective, notes that the articles in this issue of the NEJM, including a letter to the editor from the authors of the original PROWESS trial report that I did not describe, provided “insights” that may guide us in deciding which patients with severe sepsis should receive activated protein C. Insights? Yes, but not yet answers.