Zymogen-Like FXa Is An Effective Pro-Hemostatic To Reverse The Anticoagulant Effects Of Direct FXa Inhibitors

Oral anticoagulants are the mainstay of treatment for prothrombotic disorders. The emerging oral factor Xa (FXa) inhibitors, which include rivaroxaban and apixaban, have been shown to be highly effective anticoagulants in several clinical scenarios, including venous thromboembolism and non-valvular atrial fibrillation. Compared to warfarin, direct FXa inhibitors have less variable pharmacokinetics, may not require routine monitoring of coagulation parameters, and have comparable to a somewhat lower bleeding risk. Despite these advantages, no approved strategy has been developed to reverse the anticoagulant effects of these drugs in the event of life-threatening bleeding or emergent need for surgery. This represents an urgent unmet clinical need. Our group has recently developed a panel of FXa mutants that are more zymogen-like than wild-type (wt)-FXa. These “zymogen-like” FXa variants have lower activity in in vitro assays compared to wt-FXa due to impaired active site maturation. Furthermore, the variants have longer plasma half-lives (>30 minutes) in vitro compared to wt-FXa (1-2 minutes) due to diminished reactivity with antithrombin III (ATIII) and tissue factor pathway inhibitor (TFPI). Remarkably however, binding to FVa rescues the activity of these zymogen-like FXa variants and as a result they are highly effective procoagulants in vivo in the setting of hemophilia (Nat. Biotech; 2011, 29:1028-33). We hypothesized that these variants could also be effective procoagulants to overcome the effects of direct FXa inhibitors. Furthermore, since direct FXa inhibitors bind the FXa active site, we expect them to compete with ATIII and TFPI for FXa binding and prolong their half-lives. We tested both of these hypotheses in in vitro coagulation studies and in vivo hemostasis models. Rivaroxaban dose-dependently inhibited thrombin generation in thrombin generation assays (TGA) when added to normal human plasma. Specifically, 500 nM rivaroxaban, the expected therapeutic steady-state plasma concentration, decreased peak thrombin generation to ∼10% of normal, and addition of 3 nM of the FXa zymogen-like variant FXa^I16L restored peak thrombin generation to 105% of normal. Higher concentrations of rivaroxaban (2.5 µM) completely abrogated thrombin generation in this assay, but 10 nM FXa^I16L restored thrombin generation to 72% of normal under these conditions. We compared these data to results obtained with other proposed reversal strategies. Gla-domainless, catalytically inactive FXa (GD-FXa^S195A), which has been shown to reverse the effects of rivaroxaban by scavenging the inhibitor, restored thrombin generation in the presence of 500 nM rivaroxaban, but required high concentrations (1 µM; >300-fold greater than FXa^I16L) to be effective. In addition, activated prothrombin complex concentrates (FEIBA), which have been shown to have some ex vivo efficacy, were ineffective under our assay conditions. In tail-clip hemostasis studies in mice, rivaroxaban dose-dependently increased blood loss, with 50 mg/kg rivaroxaban resulting in 217% of normal blood loss. Addition of FXa^I16L (200 mg/kg) reduced rivaroxaban-induced blood loss to 141% of normal. To examine the effect of rivaroxaban on the half-life of FXa, we pre-incubated FXa^I16L or wt-FXa with or without rivaroxaban in normal human plasma and then performed TGA experiments after various incubation times. When wt-FXa or FXa^I16L were pre-incubated in plasma in the absence of rivaroxaban, their half-lives were 4.6 minutes and 1.37 hours, respectively. Remarkably, when wt-FXa or FXa^I16L were incubated in plasma in the presence of 500 nM rivaroxaban, their respective half-lives were prolonged to 9.4 hours (123-fold increase) and 18.1 hours (13.2-fold increase). These results suggest that a zymogen-like FXa variant, FXa^I16L, can reverse the effects of rivaroxaban in vitro and in vivo. Furthermore, FXa^I16L is a bypassing agent that only requires catalytic amounts of protein, in contrast to scavengers or “true” antidotes like GD-FXa^S195A that require stoichiometric concentrations. This indicates that much lower quantities of FXa^I16L may be effective in vivo. We also showed that rivaroxaban dramatically prolongs the half-life of FXa in plasma, possibly by competing with ATIII and TFPI for FXa binding. This work provides a starting point for the development of a long half-life reversal strategy for the emerging FXa inhibitors.