Sepsis editorial from JAMA (same issue) contains the following excerpt relating to AT III's failure the KyperSept trial:
...Both rAPC and antithrombin III are naturally occurring anticoagulants that produce their effect by down-regulating the activity of the coagulation cascade. Since their mechanism of action with respect to the coagulation cascade is similar, it is crucial to determine why they appear to have different efficacy in patients with sepsis. The observed differences in outcomes between the 2 studies are not likely to be due exclusively to study design or implementation issues. The probability of a systematic bias in the study by Warren et al that precluded detection of a true benefit of antithrombin III is low, and the sample size makes a type II error unlikely, suggesting that the observed differences in these 2 trials were due to a mechanistic difference between rAPC and antithrombin III. This contention is supported by the observation that there was an interaction between antithrombin III supplementation and heparin therapy. If the beneficial effect of rAPC and antithrombin III were solely related to their ability to down-regulate coagulation (thus inhibiting microvascular thrombotic complications of this disorder), patients who received heparin and antithrombin III should have an improved outcome when compared with patients receiving antithrombin III alone, or placebo, if no harmful interactions occur. The opposite was observed in the study by Warren et al, suggesting that it is not only the anticoagulant effect of these agents that is of greatest benefit in this setting. The nature of the beneficial effect of rAPC requires additional study. An attractive, but unproven, hypothesis is that rAPC down-regulates the inflammatory response to sepsis. The trial by Bernard et al suggests that the anti-inflammatory effect of rAPC might be greater than that of antithrombin III.
Does the study by Warren et al confirm that there is no role for antithrombin III supplementation in sepsis? The answer is likely yes, at least for patients similar to those enrolled in this trial. However, the utility of antithrombin III in highly selected patients has not been completely proven or disproven. For example, the authors suggest that there was a trend toward reduced long-term mortality in patients who were sickest, as reflected in elevated Simplified Acute Physiology Score version II scores. This might be expected for an agent such as an anticoagulant that targets a process that is a relatively late sequela of sepsis, but this is based on a subgroup analysis, and is a hypothesis that needs to be tested. Does severity of illness emerge as a significant effect modifier in a multivariate model? Does the course of evolution or resolution of organ dysfunction mirror physiological events that might be expected with reversal of coagulopathy? These questions can be readily answered if they are anticipated in advance, and incorporated into the design of future trials.
Warren et al also found evidence of an adverse interaction between the study medication and heparin. Since heparin is widely used in the intensive care unit to prevent venous thromboembolic disease and to maintain intravascular catheter patency, this interaction would have important implications were antithrombin III used frequently for the treatment of sepsis, as it is in some countries.14 There are also implications for other agents with antithrombotic properties such as rAPC. Is there a correlation between the dose of heparin received and the risk of bleeding complications, or change in the hematologic component of the organ dysfunction score? Was the impact of heparin therapy primarily ablation of an anti-inflammatory effect (that might have been evident, for example, by evaluating levels of an inflammatory marker such as IL-6) or did it occur because heparin amplified the potential toxicity of therapy? While major bleeding complications were significantly more common in patients receiving concomitant heparin (Table 4 of Warren et al), the RR of 1.77 in treated patients is not substantially different from the RR of major bleeding of 1.71 for treated patients who did not receive heparin, and the statistical significance of the findings in the former subgroup can be readily explained on the basis of the larger sample size.
The obstacle to discovery is the illusion of knowledge.
