We read with great interest the study by Mansour et al., which provided a longitudinal examination of plasma antithrombin levels during extracorporeal membrane oxygenation (ECMO) and investigated their correlation with heparin responsiveness.  The study offers important clinical insights for intensivists managing ECMO; however, it merits further discussion on certain clinical and technical points.

The authors’ decision to utilize Liquid anti-Xa and Stachrom ATIII assays for quantifying antithrombin and anti-Xa levels was justified, given the established anti-Xa and anti-IIa activities of heparin. Nonetheless, interpreting these results must be done with caution. Specifically, the Liquid anti-Xa assay employed does not incorporate dextran sulfate, which is typically included in commercial kits to reduce nonspecific binding of heparin to plasma proteins such as histidine-rich glycoproteins and platelet factor 4. The absence of dextran sulfate leads to biases of as much as 0.24 units/ml between anti-Xa assays.  An underestimation of heparin’s activity is possible with the Liquid anti-Xa assay, especially considering that platelet factor 4 levels may rise during ECMO or following venipuncture. Additionally, it should be noted that the Stachrom ATIII assay, which involves exogenous human thrombin, may overestimate antithrombin levels in the presence of heparin, as thrombin could be inhibited not only by antithrombin but also by heparin cofactor II. Although the effect of heparin cofactor II is likely small, it could still contribute to inter-individual variability in measured antithrombin levels, as depicted in their supplemental figure 1. 

As a clinically relevant secondary endpoint, the authors reported bleeding events in 36 patients (72%).  The definition of bleeding was broad, including common cannula site bleeding, occasional intracranial hemorrhages, and significant hemoglobin drops (greater than 2 g/dl) necessitating transfusion. Often, antithrombin levels were not accessible at the time of bleeding; therefore, the number of bleeding events was analyzed in relation to time-weighted averages of antithrombin levels. Although the authors noted an increased prevalence of bleeding in patients with time-weighted antithrombin levels above 70%, the timing and extent of bleeding were highly dependent on the clinical setting. The article’s supplemental table 4 shows that a majority of cardiotomy patients, many of whom had a low antithrombin activity (median, 45%) at the start of ECMO, experienced bleeding.

The authors highlighted the unique coagulation challenges presented by ECMO. Heparin’s anticoagulant effects impact both tissue factor and contact pathways through AT-dependent and independent mechanisms. Although antithrombin-dependent actions have been conventionally thought to be essential in preventing circuit thrombosis, antithrombin-independent heparin actions could impact both vascular hemostasis and prevention of thrombosis during ECMO. Sub-picomolar concentrations of tissue factor are considered a main coagulation trigger after vascular injury.  Heparin infusions can increase tissue factor pathway inhibitor levels by two- to four-fold which suppresses the tissue factor-Factor VIIa complex and inhibits the early-phase prothrombinase complex with protein S as a cofactor. Further, heparin can enhance fibrin porosity, increasing susceptibility to fibrinolysis. These antithrombin-independent heparin effects potentially impact bleeding diathesis without measurable changes in anti-Xa assay or partial thromboplastin time values.

In summary, we emphasize that the variation in preanalytical conditions and the distinct endpoints of the assays used to measure heparin activity might have led to discrepancies between measured values and their clinical correlations. We commend the authors for their work towards improved anticoagulation monitoring in ECMO. Nevertheless, their findings underscore the need for further studies in these patients at risk for both bleeding and thrombosis.