Bleeding complications in high-risk pulmonary embolism hospitalizations with use of extracorporeal membrane oxygenation.

Background: Extracorporeal membrane oxygenation (ECMO) offers potentially life-saving circulatory support in patients with high-risk pulmonary embolism (PE) and refractory hemodynamic instability. We sought to investigate the bleeding complications associated with ECMO in conjunction with reperfusion strategies in high-risk PE.
Methods: We identified high-risk PE hospitalizations in the US Nationwide Inpatient Sample (years 2016-2020) with use of ECMO (as stand-alone strategy or in conjunction with thrombolysis-based [systemic thrombolysis or catheter-directed thrombolysis] and mechanical [surgical embolectomy or catheter-based thrombectomy] reperfusion strategies). We investigated in-hospital outcomes related to overall bleeding, major bleeding (defined as subarachnoid or intracerebral bleeding, hemopericardium, retroperitoneal bleeding, hemoperitoneum, hemothorax, or hemarthrosis), non-traumatic intracranial hemorrhage, transfusion and vascular complications (including intra- and post-procedural injuries of vessels and organs, hemorrhages and hematomas).
Results: Among 122,735 hospitalizations for high-risk PE (51% with cardiac arrest), ECMO was used in 2,805 (2.3%); as stand-alone strategy in 1.4%, with thrombolysis-based reperfusion in 0.4%, and with mechanical reperfusion in 0.5%. In patients with ECMO, bleeding overall was reported in 63% (vs. 32% in patients without ECMO, p<0.001), major bleeding in 35% (vs. 19%, p<0.001), intracranial hemorrhage in 4.5% (vs. 2.8%, p=0.02), transfusion in 22% (vs. 15%, p<0.001) and vascular complications in 10% (vs. 1.7%, p<0.001). ECMO was associated with more overall bleedings, major bleedings, and transfusions compared with neither reperfusion nor ECMO in all patients in high-risk PE as well as in the cardiac arrest subgroup, irrespective of whether ECMO was used as stand-alone treatment or in conjunction with a reperfusion strategy (Figure). Intracranial hemorrhage was not significantly associated with any of the ECMO strategies. Vascular complications were significantly greater in the ECMO strategies compared with all other groups. Among patients with ECMO, there were no significant differences in major bleeding in thrombolysis-based reperfusion (OR 1.15, 95% CI 0.69-1.91) or mechanical reperfusion (OR 0.93, 95% CI 0.59-1.48) compared with no reperfusion.
Conclusion: In patients with high-risk PE and refractory hemodynamic instability, ECMO is associated with increased bleeding risk irrespective of whether reperfusion is performed. ECMO treatment should be provided by highly trained teams to minimize procedure associated problems and prevent bleeding complications.
Figure. Multivariable regression model for the bleeding outcomes among patients with high-risk PE (A) and cardiac arrest only (B) according to ECMO and reperfusion therapies (reference level “no reperfusion therapy or ECMO”).
ECMO: extracorporeal membrane oxygenation; PE: pulmonary embolism