In vitro models for the experimental evaluation of mechanical thrombectomy devices in acute ischemic stroke

Abstract

Mechanical thrombectomy has become an important method for the treatment of acute ischemic stroke for large vessel occlusions. The current hotspots of mechanical thrombectomy are optimizing the treatment methods, improving the recanalization rate and reducing complications. The in vitro model has become a common and convenient method for mechanical thrombectomy research. This review summarizes the in vitro model in the following aspects: the preparation of clot analogues; the experimental platform; the application of the in vitro model in the testing of thrombectomy devices; and the advantages, limitations and future trends of the in vitro experimental model. This review describes the characteristics and applications of the in vitro experimental model with the hope that the in vitro experimental model will be further improved and play a more effective role in the study of mechanical thrombectomy.

Introduction

Endovascular mechanical thrombectomy has become an important treatment for large vessel occlusion in acute ischemic stroke. The devices used for mechanical thrombectomy are stent retrievers or aspiration catheters. The current successful revascularization rate is approximately 80–90%, ¹ and recanalization is associated with good clinical results. There are still some problems with mechanical thrombectomy, such as failed recanalization, distal vessel embolization and vessel injury. Novel devices and therapeutic techniques are constantly emerging to address these problems in the field of endovascular mechanical thrombectomy. In addition to clinical studies, there is a need for faster and less costly methods to study the mechanisms of mechanical thrombectomy and complications and to evaluate new devices and methods.

Common models for mechanical thrombectomy testing are in vitro models and in vivo animal models. ² Animal models, usually including swine, ³ canine, ⁴ and rabbit ⁵ models, could provide a more realistic physiological environment for mechanical thrombectomy and allow for safety evaluation. However, animal experiments require more complex ethical requirements, and the vascular morphology in animals is different from that in humans. In contrast, in vitro experimental devices simulate the morphology of blood vessels, blood flow status, and artificial embolism in an in vitro environment, which is easy and intuitive, and are a good method for evaluating thrombectomy devices. Before the advent of stent retrievers, in vitro experimental methods were used to examine previous generations of cerebral or peripheral vessel thrombectomy devices ⁶ and study cerebral blood flow characteristics. With the advancement of interventional techniques and the widespread use of stent retrievers and aspiration catheters, the in vitro experimental model has been commonly used in the study of mechanical thrombectomy for acute ischemic stroke.

In vitro models play an important role in evaluating the performance of stent retrievers and aspiration catheters and in studying the interaction of devices with clots. Novel devices and methods for mechanical thrombectomy are also commonly tested first by in vitro models. We reviewed the articles on in vitro mechanical thrombectomy test models in recent years and found differences between the articles in the setup of in vitro experiments, including the preparation of clot analogues, the material and shape of the vascular model, and the internal fluid dynamics parameters.

This article reviews the use of in vitro experimental models in acute ischemic stroke for mechanical thrombectomy studies. In this review, we first describe the thrombus preparation for in vitro experiments and the model composition. In these sections, we review the clot analogue prepared under dynamic conditions or made by nonblood materials and pressure and flow monitoring in the fluid circuit, which have been less discussed in previous reviews. Then, we provide an overview of the application of in vitro experimental devices in the study of stent retrievers, aspiration catheters, and some new devices and methods. Finally, we discuss the advantages and limitations of current in vitro models, and we provide an outlook on future trends in this field. Through this article, we hope to gain a clearer understanding of the in vitro experimental model and make it more useful for studies on acute ischemic stroke and the evaluation of new devices and methods.

Clot analogue preparation

In vitro experiments for mechanical thrombectomy often require the preparation of clot analogues, which is also necessary for animal studies of mechanical thrombectomy and research on the characteristics of clots. Blood for the preparation of clot analogues is usually obtained from humans, ⁷ bovines, ⁸ ovines ⁹ and swine. ¹⁰

Clot analogues can be prepared in a variety of ways. Coagulation from whole blood under static conditions is a convenient method. Thrombin is commonly added to promote clotting. The clot analogues made by this method are rich in red blood cells. Gralla et al. ¹¹ used thrombin-induced clots in their animal model for mechanical thrombectomy. In their experiment, 25 IU thrombin and 1 g barium sulfate were added to 10 ml swine autologous blood. Barium sulfate allows the clots to have good radiopacity under angiography. A number of subsequent animal studies and in vitro studies are based on this method to produce clot analogues. Some studies made blood clots spontaneously without using thrombin .^12,13 Malone et al. found that the structural integrity and stiffness of clot analogues with thrombin addition is increased. ¹⁴

The proportion of red blood cells and fibrin in human clots is diverse, and fibrin-rich clots are stiffer and more challenging to retrieve. To evaluate the thrombectomy performance with different types of clots, Duffy et al. ⁹ proposed a method that controlled the amount of each component in the blood clot. Sodium citrate was immediately added to the collected ovine blood for anticoagulation. Then, the blood was centrifuged to separate the blood constituents. Different ratios of red blood cells and plasma were mixed. After adding calcium chloride and leaving it for approximately half an hour to one hour, clot analogues of varying composition ratios were formed. Clots prepared by this method, especially fibrin-rich clots, have appeared in many in vitro experiments.^13,15–19 Platelet-contracted clots ²⁰ and calcified clots ²¹ have also been made by modifying the above method.

Clots can also be prepared under dynamic conditions. The Chandler loop, first reported in 1958, is used to generate clots by simulating blood flow conditions in vitro. It contains a sealed circular plastic tube with blood in it, and the circular loop usually rotates at approximately 10–30 rounds per minute in a water bath or heated incubator to control the temperature. The Chandler loop has been used to prepare clots in several in vitro studies,^19,22–24 and histological examination showed that the red blood cells and fibrin content of the thrombi made by the Chandler loop are similar to those of humans. ²⁵ A number of other studies on the preparation of thrombi have also used dynamic conditions.^7,26,27

The composition of clot analogues has an important influence on their mechanical behaviour and the performance of thrombectomy devices.^28,29 The histological or mechanical properties of the clot analogues should be similar to those of real clots. Haematoxylin and eosin (H&E) staining and Martius scarlet blue (MSB) staining are often used to analyse the fibrin and erythrocyte components of clots. Platelet content in blood clots can be measured by immunohistochemical staining.^1,7,30 Scanning electron microscopy can be used to visualize microstructures of clot analogues.^18,20,29 There are various methods for testing the mechanical properties of clot analogues, including tensile,^{14, 27} compression,^14,29,31 creep, ¹⁴ stress relaxation, ²⁹ indentation,^14,18 rheometry ¹⁴ and friction testing. ³² Chueh et al. found that clot analogues made from bovine blood are stiffer than human and porcine blood clots. The addition of barium sulfate increases the elasticity of clot analogues. ³¹ Johnson et al. found that clot analogues showed cyclic loading–unloading hysteresis and stress relaxation behaviour. ²⁹ Thrombi prepared under dynamic conditions have longer tensile lengths and are less likely to break during catheter injection testing. ²⁷

Clot analogues prepared from nonblood materials have been used in some thrombectomy experiments. Polyurethane ³³ or a mixture of guar, borax and water ³⁴ are used to fabricate clot analogues to test aspiration catheters. An in vitro study used crustacean haemolymph ³⁵ to make clots. In an animal study, ethylene vinyl alcohol copolymer ³⁶ was used to model an ultrahard clot to test the performance of the stent retriever. A study compared the shear modulus and elasticity models of synthetic thrombi made from hydrogels of polyacrylamide and alginate and thrombi made from human blood. ³⁷ Although the mechanical properties of some clot analogues of nonblood materials have been tested,^34,37 their frangibility has not been reported, and these materials have not been applied to study distal embolization.

Experimental platforms

The in vitro model for the experimental evaluation of mechanical thrombectomy devices usually contains a fluid circuit with a vascular replica.^2,8,38 The fluid circuit is filled with saline or glycerol-saline mixture and usually warmed at 37 °C. Pulsation pumps are most often used to generate the approximate physiological flow in the fluid circuit. Simplified stationary flow has also been used in some studies. The data acquisition system can be connected to the loop to measure the pressure and flow. Similar equipment has been used in other in vitro studies to investigate cerebral blood flow in the circle of Willis ³⁹ or to analyse clot trajectories. ³⁵

Three-dimensional transparent silicone vascular replicas, which can realistically reproduce the morphology of cerebral blood vessels and have mechanical properties close to physiology,^38,40 are most widely used in in vitro experimental platforms.^10,21 Some studies have the complete circle of the Willis model,^21,41,42 while others studies only have the unilateral anterior circulation model.^8,22 Due to the high coefficient of friction of silicone, some studies added surfactant into the test fluid^10,42,43 or used lubricant coatings^8,22 on the inner wall of the vascular model to reduce the friction. During thrombectomy, the interaction of the device and the thrombus can be observed through the transparent vascular model, and the restoration of flow can be verified under angiography. ⁴¹ Glass,^1,24 PMMA ³⁴ or other 3D printed materials^38,41,44 have also been used to fabricate vascular replicas. Depending on the objectives of the study, some studies simplified the vascular model into straight, cone-shaped or curved tubes comprising soft or rigid materials.^17,45 Recently, a test bed with a cadaveric cerebrovascular connecting to a peristaltic pump has been reported. ⁴⁶

Silicone vascular models can be produced by injection moulding or 3D printing. There are many commercially available silicone vascular models with moderate cost, fine transparency, and similar mechanical properties close to vascular tissue. The rigid vascular model cannot reflect the deformation of the blood vessels, and therefore, the model has a greater degree of distortion. 3D printing is suitable for small batch production in the laboratory, but some materials are less transparent. No research has been found comparing the friction properties of several vascular model materials. For a model containing the circle of Willis, there may be more realistic pressure and flow distributions because of the inclusion of more realistic collateral circulation, while no reports have been found that compare the internal hydrodynamics of a unilateral model with that of a model containing a completed circle of Willis.

Some platforms include pressure sensors, flow sensors and data acquisition systems to monitor the flow state in the vascular replica. The pressure and flow rate in the vascular replica can be controlled in the physiological range by adjusting the pump output and flow resistance devices, such as clamps and valves installed in the fluid circuit. In some simplified situations, the reservoir is placed above the test platform and connected to the proximal side of the embolic site, creating static pressure on the clot.^16,47 Although most of the in vitro experiments set up the flow and pressure in the circuit prior to the thrombectomy test, only a few papers have reported changes in pressure and flow after embolization and during mechanical thrombectomy procedures. ⁸ After the baseline flow and pressure are set, if the output flow rate of the pump and the resistance of the loop remain unchanged, the flow state in the loop will change when the tested vessel is occluded by clots. If the flow loop lacks reasonable collateral circulation (for example, the branch where Valve 3 is located in Figure 1A), the pressure and flow may deviate significantly from the physiological situation once the tested vessel is occluded. Deviating pressures and flow rates may lead to inaccurate test results, especially for aspiration catheters. The settings of many in vitro mechanical thrombectomy studies are summarized in Table 1.

Figure 1.

Composition of the in vitro experimental platform in our study (A); thrombin-induced clot analogue (B); unilateral anterior circulation model (C); complete circle of Willis model (Preclinic Medtech, Shanghai, China) (D); a stent retriever prototype developed by our group was tested in an in vitro circulation model (E).

Table 1.

Summary of sets of in vitro thrombectomy experiments.

Researchers	Clots	Vascular replica	Test fluid	Pump	Flow rate	Pressure
Chueh 8	thrombin-induced, bovine blood	silicon, unilateral ICA/MCA, LSR topcoat	60% water, 40% glycerine	piston pump	CCA: 363  ±  1.9 mL/min MCA: 132  ±  1.4 mL/min	MCA pressure gradient: 21  ±  0.7 mmHg
Chueh 48, 49	thrombin-induced, bovine blood human blood	silicon, unilateral, with leptomeningeal collaterals and PComA	saline	peristaltic pump	MCA: 131.6  ±  3.2 mL/min PcomA: 23.3  ±  9.4 mL/min ACA: 67  ±  5.1 mL/min	physiologically representative pressure gradient
Chueh 50	thrombin-induced, bovine blood, human blood	silicon, unilateral, with leptomeningeal collaterals and PComA	saline	peristaltic roller pump	CCA: 340  ±  15.1 mL/min MCA: 131  ±  2.7 mL/min PcomA: 11.4  ±  2.7 mL/min	proximal MCA: 73.7  ±  5.2 mmHg MCA pressure gradient 5.6  ±  4.2 mmHg
Arslanian 51	thrombin-induced, bovine blood, fibrinogen and calcium-phosphate	silicone, complete circle of Willis	saline	peristaltic roller pump	/	/
Madjidyar 22	Chandler loop	silicon, unilateral vascular, coated with slippery liquid (Elastrat Sarl)	60% water, 40% glycerine, 37 °C	programmable piston pump, 60 bpm	velocity 1.2 m/s	pressure 140/80 mmHg
Madjidyar 19, 52	Chandler loop, fibrin-rich clot	silicon, unilateral vascular, coated with propranolol-based fluid (Elastrat Sarl)	60% saline, 40% glycerine, 37 °C	pulse 60/min	velocity 100 cm/s	mean pressure 100 mmHg
Machi 12	white ‘stiff’ thrombi, red ‘soft’ thrombi	rigid 3D printed unilateral anterior circulation	saline, 37 °C	/	/	/
Machi 44	organized clot human blood	3D printed, translucent photosensitive resin	water, 37 °C	gear pump	/	/
Kaneko 10	thrombin-induced, porcine blood	silicon, unilateral anterior circulation,	60% saline, 40% glycerine, detergent, 37 °C	perfusion pump	ICA: 240 mL/min, ECA: 160 mL/min	hydrostatic pressure, 100–126 mmHg
Fennel 15	fibrin-rich clot, ovine blood	near complete circle of Willis circulation	water, 37 °C	pulsatile pump	full model: 590 mL/min	/
Wiesmann 45	/	silicone, straight, curved, and curved cone-shaped tubes	saline, iodinate contrast agent, 22 °C	/	/	/
Girdhar 43	fibrin-rich, porcine blood, thrombin induced	silicone	water, 0.2% v/v soap, 37  ±  2 °C	pulsatile pump, 60 bpm	300  ±  50 mL/min	systolic 150  ±  10 mmHg diastolic 70  ±  10 mmHg
Mokin 41, 53, 54	thrombin-induced, swine blood	3D printed, TangoPlus material, near-complete circle of Willis	60% water, 40% glycerine, 37 °C	pulsatile pump	/	/
Lally 47	naturally clotted, ovine blood	glass, pinch point	water containing haematoxylin particles	fluid reservoir 122 cm above the model	/	mean arterial pressure 90 mmHg
Nogueira 55 Fitzgerald 42	red-blood cell-rich, mixed, fibrin/platelet-rich	3D printed, circle of Willis	1:4 glycerol:water, surfactant, 37 °C	peristaltic pump, 72 bpm	750 ml/min	average pressure gradient ICA 89 mmHg, M1 112 mmHg
Soize 13	red-blood-cell-rich, fibrin-rich, hybrid, ovine blood	silicone, unilateral anterior circulation (Elastrat), posterior flow	saline solution, 37 °C	pulsatile pump	430 ml/min	110/60 mmHg
Rizvi 16	RBC-rich, fibrin-rich, mixed, rabbit blood	glass model, pinch-point	water	fluid reservoir 122 cm above the model	/	mean arterial pressure 90 mmHg
Ohshima 56	white hard clot, red soft clot, swine blood	silicone, unilateral anterior circulation	saline, 37 °C	/	main route: 50 cm/s collateral route: 10 cm/s	/
Luraghi 17	ca 20% RBC	glass U-bent vessel, silicon funnel-shaped vessel, 3D-printed silicone circle of Willis	saline 37 °C	stationary flow	/	/
Wenger 24	Chandler loop, porcine blood	glass model	water 37 °C	pulsatile pump, 50 bpm	medium flow 125 mL/min	pulsatile pressure 110–150 mmHg
Bernava 34	mixtures of guar, borax and water	carved PMMA model	/	/	/	/
Reddy 38	elastic emboli, fragment-prone emboli	3D-printed clear resin, glass, silicone, unilateral anterior circulation	saline solution, replicator fluid	syringe pump, 60 bpm	ICA: 265  ±  62 ml/min MCA: 121  ±  28 ml/min	calibrated to clinical records
Liu 1	mixing plasma and red blood cells 9:1	glass, unilateral anterior circulation	/	peristaltic pump	physiological flow	physiological pressure
Fargen 57	surrogate material	silicone, unilateral anterior circulation (Elastrat)	saline	dual output solenoid diaphragm pump	120 mL/min	120/80 mm Hg
Johnson 21	calcified cerebral emboli	silicone, complete circle of Willis (Elastrat Sàrl)	saline, 37 °C	continuous flow pump	full model: 680  ±  50 ml/min ICA: 287  ±  35 mL/min	/

Applications in testing thrombectomy devices

The in vitro model can be used to evaluate the performance of thrombectomy devices, such as recanalization rate and time, number of recanalization manoeuvres, distal embolization^22,48,56 and first pass success rate. ⁴³ Soft and hard clots ¹² and vascular models with various tortuosities^10,58 can be used to assess the performance of thrombectomy devices in a variety of situations. For stent retrievers, in addition to the items mentioned above, in vitro models can also test wall apposition^12,23 and tensile force.^12,23 Some new segmented, ¹⁰ and longer ⁴³ stent retrievers designed for capturing hard clots ¹⁵ are being tested through in vitro models. For aspiration catheters, in vitro models have been used to evaluate the effect of aspiration distance,^16,47 aspiration angle, ³⁴ catheter diameter ⁵⁵ and cyclical aspiration.^33,51,59 The thrombectomy test of the combined use of stent retrievers and aspiration catheters^22,41 or the combined use of balloon guide catheters and thrombectomy devices^52,56 in vascular replicas has also been reported. Some studies compared the recanalization rate of stent retrievers and aspiration catheters through in vitro platforms ⁵³ and the difference in effectiveness of these two types of devices for different types of clots.^19,20

The in vitro experimental model provides a convenient method for testing new devices and new technologies emerging in the neurointervention field. A braided mesh device designed to protect distal embolization lead by clot fragments was tested in in vitro models.^49,54,57 Some large bare or funnel-shaped catheters ⁶⁰ also showed good results in an in vitro environment. In vitro models are suitable tools for validating the performance of intracranial neurovascular intervention robotic systems. ⁶¹ In a preclinical study, a magnetically manipulated robotic neurointerventional platform assisted mechanical thrombectomy in an in vitro cerebral vascular phantom. ⁶² In terms of thrombectomy methods, Wiesmanns et al. ⁴⁵ used an in vitro model to evaluate the role of active pushing techniques on the effectiveness of recanalization. Fast stent retrieval was found to increase the recanalization rate in an in vitro flow model. ¹³

Advantages of in vitro models

Study of distal embolization

Distal embolism caused by dislodged clot fragments is a complication of mechanical thrombectomy and may affect the outcome of the treatment. One of the advantages of in vitro models is the ability to collect fluid in the circulating line, which can be used to evaluate the size and the number of thrombus fragments during thrombectomy and to reflect the extent of distal embolism. Using a particle size analyser based on the Coulter principle, Chueh et al. analysed the size and number of emboli caused by several different devices and different treatments.^{8,48–51, 63} Because the vascular model used by Mokin ⁴¹ contained smaller diameter vessels, their study used angiography and TICI scores to assess the degree of recanalization. A filter deployed at the distal end of the vessel replica can also be used to collect emboli.^42,52,64

Investigating the interaction of devices and clots

Studying the interaction between the stent retriever and the thrombus is important for the optimal design of a stent retriever. Despite the increased radiopaque techniques to some stent retrievers, it is difficult to observe the deformation of the stent retriever and the thrombus under angiography. The in vitro experimental platform allows visualization of stent–thrombus interactions when using transparent vessel models. Madjidyar et al. ²² observed that the thrombus was compressed aside by the stent retriever and that the thrombus rolled distally during retrieval. Machi et al. ¹² found that for large volumes of fibrin-rich thrombi, the stent retriever may not be able to expand, resulting in failure to recanalize. Liu et al. ¹ observed the interaction of the device with clots in the ADAPT and CAPTIVE techniques in an in vitro model.

The interaction of a stent with a thrombus can be further observed by 3D image reconstruction. The deformation of the stent retriever and thrombi in a straight silicone tube was detected by flat panel computed tomography, showing a cushion-like bulging thrombus into the stent lumen. ⁶⁵ Marel et al. ⁶⁶ proposed the concept of clot integration factor and observed the internal volume of soft and hard thrombi embedded in the stent by high-resolution cone beam CT. Micro computed tomography is also used to evaluate strut indentation. ¹⁸

Verification of simulation models

Mechanical thrombectomy procedures can be virtually simulated by finite element analysis. The deformation and stress of devices, clots and vessels can be well predicted in the simulation. Finite element analysis has potential in the development of novel thrombectomy devices and the evaluation of new surgical methods. Using 3D reconstruction technology and 3D printing technology, the morphology of the in vitro vascular model and the in silico vascular model are consistent, which is favourable for the validation of the simulation model. Luraghi et al. ¹⁷ established a simulation process for stent retriever thrombectomy based on the finite element method and performed in vitro thrombectomy tests and simulation thrombectomy tests under three conditions: glass U-shaped tube, silicone funnel-shaped tube, and 3D printed silicone vessel. The results of the in vitro tests were found to be consistent with the simulation results, and the accuracy and reliability of the simulation model were confirmed. Computational fluid dynamics can be used for fluid analysis in aspiration. Lally et al. ⁴⁷ verified the flow direction of fluid at the catheter tip in the simulation using an in vitro model.

Limitations and future prospects

There are still some limitations in the in vitro model for mechanical thrombectomy. First, the vascular replica is still far from the real cerebral vasculature in detail. If there is a lack of lubrication, the silicone vascular model has a larger coefficient of friction. The rigid vascular model, on the other hand, does not reflect vascular traction deformation during thrombectomy. Vascular injury cannot be assessed using in vitro models, and in vivo studies are needed. Most of the current vascular models only extend to the M2 segment, and some of the studied vascular models can reach the M4 segment. ⁴¹ However, the level of detail of the models is still not comparable with that of real cerebral vessels. Second, although the viscosity of the simulated fluid can be set to approximate that of blood, the in vitro model cannot simulate the natural generation and degradation of thrombi and the action of thrombolytic agents due to the lack of active substances. Third, many studies did not check the pressure and flow rate in the circuit during thrombectomy tests, which may lead to inaccurate test results if the pressure and flow rate in the circuit are too different from the physiological situation. In general, there is still room for improvement in many aspects of the in vitro model.

The in vitro models for thrombectomy device testing may be integrated with more novel technologies in the future. Some features, such as atherosclerosis or calcified plaque models, may be added into the vascular models. Tissue engineering and 3D bioprinting technology have made it possible to fabricate bioactive artificial blood vessel models. Microchannel networks fabricated with photolithography techniques have been used in microfluidic oxygenators, and they have the potential to be applied to vascular models to simulate distal cerebral vessels with smaller diameters or perforating arteries. Future in vitro models may integrate more microsensors in the flow circuit, for example, for pressure and temperature monitoring. In the future, there may also be new approaches to preparing artificial clots that more closely resemble the in vivo clots in terms of tissue composition and mechanical properties.

Conclusion

The in vitro experimental model has the advantages of simplicity and intuitiveness and is widely used in the study of interventional mechanical thrombectomy. An in vitro model can be used to evaluate stent retrievers, aspiration catheters and some new thrombectomy devices and methods. The in vitro experimental model also has the unique role of being able to study the number of emboli fragments, evaluate the interaction between the device and the thrombus, and validate the simulation model. Although the current in vitro model has some limitations, the in vitro model may play a more important role in the treatment of acute ischemic stroke with the development of experimental technology.