We thank Martucci et al. for their letter and their insightful remarks on our article.  Their comments highlight the importance of a close collaboration between anesthesiologist-intensivists and hemostasis specialists, as well as a shared understanding of laboratory coagulation tests. They also provide us with the opportunity to clarify the critical issue of measuring anti-Xa levels.

Indeed, not all anti-Xa assays are equivalent. One point is that most of them, but not all, contain dextran sulfate, the characteristics and concentration of which may vary by manufacturer. The intended use of added dextran sulfate is to displace unfractionated heparin from platelet factor 4 released in the blood collection tube by platelets, i.e., to recover circulating active unfractionated heparin. However, this may not always be fully effective. To our knowledge, the clinical impact of choosing one anti-Xa reagent over another has not been evaluated, and this requires well-designed studies.

However, we respectfully differ from Martucci et al. on two key points:

  1. Dextran sulfate not only displaces unfractionated heparin from complexes formed in vivo with various proteins (unfractionated heparin interactome), but also from circulating protamine, which is used to neutralize unfractionated heparin after cardiac surgery. In such cases, affecting half of the patients in our study at early time-points, using dextran sulfate would have significantly overestimated anti-Xa levels.
  2. As Martucci et al. noted, platelet factor 4 can also be released in vivo during acute intensive care unit situations, particularly with venoarterial extracorporeal membrane oxygenation (VA-ECMO). In such scenarios, dextran sulfate could cause in vitro dissociation of in vivo-formed complexes, in which unfractionated heparin is inactive, leading to an overestimation of anti-Xa levels as well.

Martucci et al. also correctly pointed out that unfractionated heparin’s antithrombotic effects may involve mechanisms independent of antithrombin, such as those mediated through heparin cofactor II and by tissue factor pathway inhibitor mobilized from the endothelium to the blood.  This leads to acknowledge that the anti-Xa assay captures only one aspect of unfractionated heparin anticoagulant effect. In our opinion, however, there is no better alternative – activated partial thromboplastin time being plagues with many drawbacks. Furthermore, regarding the specificity of the functional assay we used to measure antithrombin, we acknowledge that the accuracy of the antithrombin assays must be taken into account.

Finally, while our study was not designed to assess the impact of low antithrombin levels on bleeding, we acknowledge that our definition of bleeding could have been improved by using established classifications, such as the Bleeding Academic Research Consortium (BARC) classification.

Overall, there is a lack of evidence to evaluate the impact of anticoagulation on bleeding and thrombotic risks during VA-ECMO support and to define an optimal antithrombotic strategy. The numerous confounding biases and competing risks make it nearly impossible, in our view, to draw causal conclusions from observational data. Therefore, we believe that the critical illness of patients requiring VA-ECMO, and the complexity of management justifies the need for additional prospective randomized controlled trials to evaluate different anticoagulation strategies in this setting.