Multimodality Quantification of Thrombus Deposition in Extracorporeal Membrane Oxygenation (ECMO); Correlating Oxygenator Computed Tomography Imaging, Electron Microscopy and Histology to Clinical Outcomes

Introduction: Extracorporeal membrane oxygenation (ECMO) is essential for supporting cardiopulmonary function in critically ill patients, However, ECMO incurs thrombogenic risks due to activation of coagulation factors and blood cells by the biomaterial surfaces. This leads to complications such as frequent device replacement and systemic embolization despite the use of anticoagulation, suggesting additional mechanisms contribute to medical device-associated thrombosis. We propose that the geometry of the ECMO oxygenator significantly influences the structure and extent of clot formation and stability. To investigate this, we employed quantitative imaging and histologic techniques for a high-resolution, three-dimensional analysis of clot morphology in ECMO components, aiming to enhance our understanding of clot formation mechanisms, correlate these observations to clinical outcomes, and develop effective mitigation strategies.

Methods: We developed a prospective study in which patients consented to the collection of blood samples, used ECMO components, and collection of relevant clinical variables. Computed tomography (CT) images were acquired using an Inveon microCT. H&E stains, scanning electron microscopy (SEM), and CT images were processed through the Dragonfly software to accomplish a 3D visualization and analysis of clot composition. Clinical demographics and ECMO characteristics were recorded for each patient.

Results: Digital photographs of used oxygenators are displayed in figure (A). Through CT scans, we quantified clot formation of distinct densities across the layers of the ECMO oxygenator (D). The results were able to quantify intraoxygenator thrombosis with high fidelity (27.65%, 14.56%, and 26.35% of the available membrane area was thrombosed across three representative oxygenators). Using a frontal CT scan to investigate clot formation identified five layers of varying structures across the intended path of blood flow (B). Clot formation was observed in layer 4 and layer 5, in which thrombosis occluded up to 12.2% of the layer on average. The overall ranking of clot formation under hemodynamic performance changes for all five layers was as follows: layer 5 > layer 4 > layer 1 = layer 2 >> layer 3. The use of anticoagulation at the time of ECMO decannulation significantly reduced the content of clot formation (P<0.0001).

We then correlated digital biomarkers from imaging with thrombus composition. Detailed histological examination (HE) and SEM was undertaken on dismantled oxygenators (B & C). We focused on thrombi at the interfaces and within the hollow fibers of the oxygenator. SEM and HE staining illustrated increased thickness, length, and density of fibrin structures in the thrombi from layer 5 to layer 4, whereas the numbers of platelets and cells decreased. Overall, layer 5 exhibited the most diverse types of thrombi, including blood cell-rich, platelet-rich, and fibrin-rich clots. We also observed aligned clots and significant clot formation in the hollow fiber area of layer 4, reflecting the geometry of the specific layer.

Discussion: This pilot study examined the potential for imaging-based modalities to evaluate device associated thrombosis ex vivo, correlating these imaging findings to direct observations of thrombus composition and location. This methodology provides a platform to investigate ECMO device failure, identifying the key locations within select oxygenators where local device modification may be of most benefit. Future research will prioritize designing membrane oxygenator properties to minimize thrombus deposition, and will allow us to incorporate the use of serial imaging endpoints in future clinical trials evaluating medical device-associated thrombosis.