“… their findings … may have clinical implications for understanding ECMO-related bleeding. While they did not conclusively demonstrate a causal relationship, their data suggest that patients who bled had significantly elevated signals of contact activation based on biomarkers.”

Blood surface interfacing that occurs within the extracorporeal membrane oxygenation (ECMO) circuit initiates contact activation, an important hemostatic pathway initiated by the Hageman factor, also called factor XII. Factor XII was first discovered in 1968, when Mr. Hageman, the index patient for factor XII deficiency, fractured his left hemi-pelvis after falling from a ladder.  On the twelfth hospital day, he suffered a massive pulmonary embolism and died. These episodes and current research suggest that Hageman factor or factor XII is not essential for intravascular coagulation, but its main roles lie in other physiologic processes.  Activated factor XII is an essential protease in the kinin–kallikrein system. When plasma is exposed to nonendothelial surfaces, like the tubing in an ECMO circuit, factor XII becomes activated, initiating kallikrein activation and subsequent cleavage of high-molecular-weight kininogen, ultimately releasing bradykinin. The role that these mediators play in thrombo-inflammatory pathways in ECMO patients remains an area with limited research.

In this issue of Anesthesiology, Helms et al. address the issue of bradykinin release and its potential for causing bleeding in a pilot study of heparin-anticoagulated ECMO patients, proposing fibrinolysis as a potential contributor to hemorrhagic complications.  Despite the small sample size (n = 30), their study presents data on the longitudinal dynamics of selected activation markers of the kinin–kallikrein system and fibrinolytic pathways in patients who experienced hemorrhage. They postulate that bradykinin release, due to kinin–kallikrein activation during ECMO, increases systemic levels of tissue plasminogen activator (tPA), which in turn enhances fibrinolysis, as assessed by levels of plasmin–antiplasmin and D-dimer (fig. 1). In patients experiencing bleeding, the researchers observed elevated levels of tPA, along with incremental changes in plasmin–antiplasmin and D-dimer levels that preceded hemorrhagic events. Although bleeding events were more frequent in venoarterial ECMO cases, statistical analyses did not indicate increased odds of hemorrhage related to anticoagulation (anti-Xa levels and activated partial thromboplastin time) or platelet counts. This important work by Helms et al. proposes a provocative hypothesis for ECMO-induced coagulopathy. However, several important points are worth discussing before we consider bradykinin, tPA, or D-dimer as a clinical predictor of hemorrhage during ECMO.

Fig. 1.

Contact activation during extracorporeal membrane oxygenation (ECMO) and its possible relation to vascular hemorrhage. Factor XII (FXII), high-molecular-weight kininogen (HK) bound to prekallikrein, and factor XI (FXI) are adsorbed onto the surface of ECMO circuits. Contact activation on the ECMO surface triggers the activation of FXII (FXIIa), which subsequently activates FXI to FXIa and converts prekallikrein to kallikrein. Other pathophysiologic events, including hemolysis of erythrocytes, formation of neutrophil extracellular traps (NETs) by neutrophils, and platelet activation, can also contribute to the activation of FXIIa. FXIa and platelets support thrombus formation on ECMO circuits, including the oxygenator membrane. Kallikrein increases bradykinin (BK) levels in circulation from high-molecular-weight kininogen. Bradykinin is a potent stimulator of tissue plasminogen activator (tPA) release via endothelial BK receptor activation. While tPA can promote thrombolysis within the ECMO circuit, as evidenced by elevated D-dimers, it can also break down the hemostatic fibrin clot at sites of vascular injury, potentially increasing the risk of bleeding. In addition, pro-urokinase (pro-uPA) is activated by kallikrein to urokinase (uPA), contributing to fibrinolysis.

Contact activation during extracorporeal membrane oxygenation (ECMO) and its possible relation to vascular hemorrhage. Factor XII (FXII), high-molecular-weight kininogen (HK) bound to prekallikrein, and factor XI (FXI) are adsorbed onto the surface of ECMO circuits. Contact activation on the ECMO surface triggers the activation of FXII (FXIIa), which subsequently activates FXI to FXIa and converts prekallikrein to kallikrein. Other pathophysiologic events, including hemolysis of erythrocytes, formation of neutrophil extracellular traps (NETs) by neutrophils, and platelet activation, can also contribute to the activation of FXIIa. FXIa and platelets support thrombus formation on ECMO circuits, including the oxygenator membrane. Kallikrein increases bradykinin (BK) levels in circulation from high-molecular-weight kininogen. Bradykinin is a potent stimulator of tissue plasminogen activator (tPA) release via endothelial BK receptor activation. While tPA can promote thrombolysis within the ECMO circuit, as evidenced by elevated D-dimers, it can also break down the hemostatic fibrin clot at sites of vascular injury, potentially increasing the risk of bleeding. In addition, pro-urokinase (pro-uPA) is activated by kallikrein to urokinase (uPA), contributing to fibrinolysis.

First, the premise of their theory depends on accurate measurement of bradykinin. Although bradykinin is considered the most potent stimulus for endothelial tPA release, its measurement is difficult due to rapid degradation in seconds and preanalytical in vitro bradykinin formation inside the sample tube. Initially undetectable levels of bradykinin may become elevated in vitro to high picogram-per-milliliter to low nanogram-per-milliliter levels from bradykinin formation in standard EDTA tubes.  Assuming all samples were collected in the same way, preanalytical bias should be evenly distributed across measurements. However, the clinical relevance of elevated bradykinin as a predictor of hyperfibrinolysis remains unclear, given the large broad CI for its association with bleeding risk (odds ratio, 0.59; 95% CI, 0.15 to 2.35). 

Second, Helms et al. reported overt bleeding incidents, mostly at the cannulation sites, following spikes in tPA and D-dimer levels. These observations support a potential causal link between fibrinolysis and bleeding; however, inherent study limitations including the assay methods preclude definitive conclusions. They suggest that D-dimer elevation is attributable to fibrin breakdown at bleeding sites, but D-dimer could also be derived from subclinical thrombus formation and fibrinolysis within the ECMO circuit, which is common and can lead to unanticipated ECMO circuit exchange.  Indeed, their results showed wide fluctuations in D-dimer levels among patients who bled, particularly after the occurrence of bleeding. Similar variations were also noted in patients without bleeding. In this small cohort (n = 30), it is difficult to determine whether contact activation and subsequent fibrinolysis occurred during ECMO randomly or continuously. It is widely assumed that contact activation peaks during the initial ECMO circuit exposure, declining thereafter as proteins are adsorbed onto the oxygenator and circuitry. 

Finally, ECMO-induced coagulopathy is multifactorial, and no single pathway alone can fully explain patients’ propensity for bleeding. Disturbances in primary hemostasis including acquired von Willebrand syndrome and acquired Bernard-Soulier syndrome are also well described when blood is exposed to supraphysiologic shear stress.  ECMO patients are also frequently thrombocytopenic and have impaired platelet aggregation in response to multiple agonists, as well as slowed initiation of thrombin generation. 

The medical history of Mr. Hageman and the findings of the current study compel us to consider whether patients with factor XII deficiency or inhibition might have reduced contact activation, decreased bradykinin release during ECMO, and/or decreased bleeding related to hyperfibrinolysis. Although we do not yet have a conclusive answer, preliminary data provide valuable insights into the complex interplay between factor XII deficiency and ECMO-related bleeding risks.

In factor XII–knockout mice, exposure of factor XII–deficient plasma to a membrane filter reduced bradykinin formation compared to normal plasma by 250-fold. This underscores the significant role of the factor XII–dependent pathway in contact-activated bradykinin generation. However, it is important to note that factor XII–independent proteases, such as endothelial prolylcarboxypeptidase, also contribute to kallikrein activation and subsequent bradykinin production. Prolylcarboxypeptidase is crucial for angiogenesis and vascular repair  particularly relevant in ECMO settings where patients tend to have extensive organ damage. Elevated bradykinin levels in these cases could indicate both ECMO-induced contact activation and extravascular healing processes.

Regarding anticoagulation for ECMO, heparin is the current mainstay agent, but its antithrombotic activity against contact activation is suboptimal as evidenced by subclinical and overt circuit thromboses. Although Helms et al. did not find any significant difference in heparin anti-Xa activities between bleeders and nonbleeders, heparin may have a role in subclinical coagulopathy, fibrinolysis, and bleeding. In support of the critical role of contact activation, Tweddel et al. reported that compared to heparin anticoagulation, factor XII depletion or factor XI depletion significantly enhanced ECMO circuit lifespan and reduced fibrinogen consumption.  Additionally, circulating tissue factor levels decreased during ECMO when factor XII or XI activity was inhibited. Notably, in lung tissues, animals treated with heparin and those with factor XI depletion exhibited moderate interstitial edema and hemorrhage, whereas factor XII depletion substantially reduced these pathologic changes. While data from ECMO in previously healthy animals may not directly translate to critically ill human subjects, these experimental data collectively indicate the possibility of enhancing ECMO safety and organ protection in future clinical trials by modulating factor XII or another component of the contact pathway.

We commend Helms et al. for their findings that may have clinical implications for understanding ECMO-related bleeding. While they did not conclusively demonstrate a causal relationship, their data suggest that patients who bled had significantly elevated signals of contact activation based on biomarkers. Emerging data compel us to further examine contact pathway inhibition for organ protection and the efficacy of anticoagulation. Mr. Hageman’s diagnosis and the role of contact activation provide a frontier worthy of scientific exploration.