High-dose treatment with purified as well as recombinant AT protected baboons challenged with lethal doses of E coli from death.72,94 Moreover, a significant reduction of the inflammatory response by AT was demonstrated.94In humans, trials95–99 investigating the effect of AT in sepsis revealed promising results. In a phase II study,95 either placebo or AT (loading dose, 90 to 120 U/kg body weight followed by 90 to 120 U/kg body weight over 4 days) was administered to patients with septic shock and DIC. AT-treated patients showed a benefit in 28-day mortality as well as a reduced duration of DIC. In that study,95high AT levels were achieved after 3 h on AT substitution. A 14-day high-dose, activity adapted, phase II study96–97 demonstrated AT at continuous AT levels > 120% to be beneficial on organ dysfunction, to attenuate DIC as well as the inflammatory response. A nonsignificant reduction of mortality as well as a shortening of the stay on the ICU was achieved in a prospective randomized study on administration of AT (loading dose, 3,000 IU followed by 1,500 IU q12h for 5 days).98 Interestingly, only modest AT plasma concentrations on day 1 and day 6 (approximately 55 to 60% and > 70%, respectively) after AT substitution were achieved.98A trial including 120 critically ill patients did not show a benefit of AT substitution (loading dose, 4,000 IU followed by 2,000 IU q12h) on survival. However, subgroup analysis in septic shock patients demonstrated a significant reduction in mortality when treated with AT.99 In that study,99 AT levels at approximately 100% could be measured immediately and 24 h after AT administration, respectively. Although the three larger studies95,98–99 were comparable with regards to the AT levels on inclusion (40 to 55%), AT levels immediately after substitution differed apparently. Therefore, it can be concluded that high AT levels should be achieved quickly after starting AT therapy.95,98–99 This is in accordance to animal models, in which the benefit of AT substitution was restricted to animals in which AT levels were achieved very early in the septic process.72 Based on these promising results, a large, double-blind, placebo-controlled multicenter trial (High-Dose Antithrombin III in Severe Sepsis [KyberSept] trial)14 investigated the effect of AT in sepsis patients. Either AT or placebo were infused to 2,314 patients (loading dose, 6,000 IU followed by 6,000 IU/d over 4 days). Heparin in prophylactic doses (≤ 10,000 IU/d) as well heparin flush for catheter patency were allowed in the study protocol. However, although considerable AT levels (180%) 24 h after AT administration had been achieved, no difference in mortality between the treatment group and the placebo group was found. Moreover, patients treated with AT had significantly more bleeding complications as compared to the placebo group.14 Subgroup analyses revealed some important clues. First, in the group of patients with AT activity levels > 60%, a beneficial effect on 90-day mortality was found, which can be explained by the higher AT activity levels achieved by AT administration in these patients, as compared to patients starting with levels < 60%. However, one could also argue that since low levels of AT predict a poor outcome, that patients starting with low levels have higher probability for fatal outcome. Second, in the subgroup analysis, patients without heparin in the AT group (n = 698) showed a clear trend toward absolute reduction of mortality with a relative mortality risk reduction of approximately 15% being significant on day 90. The interpretation of these results is difficult. A probable explanation for that finding might be that concomitant use of heparin decreases the ability of AT to bind to glycosaminoglycans on endothelial cells.100–101 That is supported by the findings in animal sepsis that administration of heparin in prophylactic doses might abolish the antiinflammatory effects of AT, such as inhibition of leukocyte-endothelium interaction or improvement of microcirculation.102 Thus, heparin may have interfered with targeting of AT to the endothelial cells. Moreover, patients treated with AT receiving no heparin had fewer bleeding complications as compared to patients receiving heparin. These results emphasized that the effect of heparin in sepsis patients is not well and properly studied. Therefore, AT substitution as a first-line therapy in sepsis patients concomitantly receiving heparin is not indicated. Whether AT substitution should be recommended in patients with sepsis not receiving heparin should be investigated in future studies.