Introduction Cerebral artery phantoms are valuable tools to test recanalization strategies of large vessel occlusion (LVO). However, these artificial models do not mimic the complex angioarchitecture and hemodynamic conditions of the human cerebrovasculature, the embolus/endothelium interaction, nor the response of delicate arterial walls to mechanical forces. In-vivo animal models are also available; however, such models do not represent the geometry, structure, or hemodynamic conditions of human cerebral arteries. To overcome these inadequacies, we present a test bed consisting of pressurized human brains which was developed and validated for LVO and revascularization.

Materials and Methods Twenty-four fresh human brains were harvested from autopsies. Internal carotid arteries and vertebral arteries were cannulated with 8F sheaths connected in parallel to a hydraulic system. Saline solution was infused at physiological flow rate and the pressure adjusted through an escape mechanism. Then, three types of representative embolus analogs (EAs) were fabricated (elastic, fragment-prone, and stiff) using a multilinear regression model derived from histology and characterization of tensile properties (including stiffness, ultimate tensile strain and stress) of sixteen emboli causing LVO strokes. EAs were introduced into the hydraulic system of the brain test bed and physiological flow was allowed to embolize the EA downstream into the cerebral vasculature to recreate LVO. Then, recanalization was attempted in 61 cases using ADAPT technique employing suction catheter (ACE™ 68; Penumbra) and in 44 cases using CAPTIVE with a stent retriever (Solitaire™ Platinum, Medtronic) and a suction catheter. Two cameras recorded videos of the embolization process and mechanical thrombectomy. Recanalization findings were used to derive into an adjusted Thrombolysis in Cerebral Infarction (aTICI) as a proxy for the modified TICI scale.

Results The test bed was highly realistic and allowed performance and direct trans-mural visualization with real-time mechanical thrombectomy for LVO. Physiological pressure with pulsatile waveforms was generated consistently by adjusting an escape valve. EAs fabricated appropriately represented the spectrum of mechanical tensile features and histology encountered in emboli removed during mechanical thrombectomy in LVO stroke. The optically semi-transparent arterial walls enabled conventional cameras to visualize the embolus-device interaction at high resolution without radiation. EAs lodged at bifurcating points of the main parent vessel with good tolerance to the physiologic flow without fragmentation and downstream migration. EAs were also observed to protrude into smaller branches and perforating arteries. We were able to successfully replicate 105 LVO cases (51 in the anterior circulation, and 54 in the posterior circulation). First pass (45%), successful (71%) and complete (60%) recanalization rates in this model by the proposed aTICI were consistent with the published results for mTICI with same devices and techniques. Direct observation of the thrombectomy procedure demonstrated that current technologies load the artery and the embolus with tensile forces, where the emboli elongate and undergo uncontrolled multifocal fragmentation leading to recurrent and residual LVO, requiring repeated recanalization attempts.

Conclusion A test bed for embolic occlusion of cerebrovascular arteries of human brains with different EAs was developed and validated for mechanical thrombectomy research and technology testing.

Disclosures L. Savastano: None. Y. Liu: None. D. Gebrezgiabhier: None. D. Evan: None. A. Reddy: None. Y. Zheng: None. A. Shih: None. A. Pandey: None.