Article Text

Download PDFPDF

Original research
A canine model of mechanical thrombectomy in stroke
  1. Olivia W Brooks1,2,
  2. Robert M King1,3,
  3. Erez Nossek4,
  4. Miklos Marosfoi1,
  5. Jildaz Caroff1,5,
  6. Ju-Yu Chueh1,
  7. Ajit S Puri1,
  8. Matthew J Gounis1
  1. 1 New England Center for Stroke Research, Department of Radiology, University of Massachusetts Medical School, Worcester, Massachusetts, USA
  2. 2 St George’s University School of Medicine, Grenada, West Indies
  3. 3 Department of Biomedical Engineering, Worcester Polytechnic Institute, Worcester, Massachusetts, USA
  4. 4 Department of Neurosurgery, NYU School of Medicine, New York City, New York, USA
  5. 5 Department of Interventional Neuroradiology, NEURI Center, Bicêtre Hospital, Le Kremlin- Bicêtre, France
  1. Correspondence to Dr Matthew J Gounis, New England Center for Stroke Research, Department of Radiology, University of Massachusetts Medical School, Worcester, MA 01655, USA; matthew.gounis{at}


Purpose To develop a preclinical model of stroke with a large vessel occlusion treated with mechanical thrombectomy.

Materials and methods An ischemic stroke model was created in dogs by the introduction of an autologous clot into the middle cerebral artery (MCA). A microcatheter was navigated to the clot and a stent retriever thrombectomy was performed with the goal to achieve Thrombolysis in Cerebral Ischemia (TICI) 2b/3 reperfusion. Perfusion and diffusion MRI was acquired after clot placement and following thrombectomy to monitor the progression of restricted diffusion as well as changes in ischemia as a result of mechanical thrombectomy. Post-mortem histology was done to confirm MCA territory infarct volume.

Results Initial MCA occlusion with TICI 0 flow was documented in all six hound-cross dogs entered into the study. TICI 2b/3 revascularization was achieved with one thrombectomy pass in four of six animals (67%). Intra-procedural events including clot autolysis leading to spontaneous revascularization (n=1) and unresolved vasospasm (n=1) accounted for thrombectomy failure. In one case, iatrogenic trauma during microcatheter navigation resulted in a direct arteriovenous fistula at the level of the cavernous carotid. Analysis of MRI indicated that a volume of tissue from the initial perfusion deficit was spared with reperfusion following thrombectomy, and there was also a volume of tissue that infarcted between MRI and ultimate recanalization.

Conclusion We describe a large animal stroke model in which mechanical thrombectomy can be performed. This model may facilitate, in a preclinical setting, optimization of complex multimodal stroke treatment paradigms for clinical translation.

  • stroke
  • intervention
  • thrombectomy

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


In 2015, mechanical thrombectomy (MT) was shown to be an effective treatment for acute ischemic stroke in patients with an emergent large vessel occlusion.1–5 Currently, the safety and efficacy of MT devices are tested in large animal models—namely, swine—with extracranial large vessel occlusions. These models have been useful to develop MT technology, but the development of multimodal stroke treatments6 requires a model of MT that includes cerebral ischemia and infarction. Importantly, since MT offers a reliable means by which to recanalize an occluded emergent large vessel occlusion, it reinvigorates the potential of neuroprotective agents that have previously failed clinical trial.6 Finding methods to preserve tissue at risk of infarction is especially important to have more time to get patients to thrombectomy-capable centers, since time is known to be a critical variable for treatment effectiveness in most patients.7 A stroke model to test MT devices, the current standard of care endovascular treatment, with adjunctive neuroprotection strategies could facilitate clinical translation.

In this study we seek to refine a previously developed canine stroke model (reviewed by Mehra et al 8) to include the introduction of a single autologous clot to create a reproducible occlusion of the middle cerebral artery (MCA), achieve microcatheter navigation to the clot, and perform stent retriever thrombectomy with a positive outcome, considered to be Thrombolysis in Cerebral Ischemia (TICI) 2b/3 recanalization.9 Additionally, the model must be amenable to MRI, which allows for assessment of the perfusion deficit and infarct evolution.


Canine model of embolic MCA occlusion

Six dogs (species: hound-cross; sex: F; age range 5.7–8.9 months; mean age 7.1 months) were used for this study. All procedures were approved by the Institutional Animal Care and Use Committee. On the day of the procedure, the animals were weighed (range 14.7–23.4 kg; mean 19.4 kg), sedated, and pretreated with intramuscular injections of a single dose of buprenorphine (0.02 mg/kg), acepromazine (0.06 mg/kg), and glycopyrrolate (0.01 mg/kg). Anesthesia was induced with an intravenous injection of propofol (3–4 mg/kg) and the animal was intubated. Anesthesia was maintained by mechanical ventilation of 1–3% isoflurane in air (tidal volume 8–10 mL/kg; peak airway pressure <20 cm H2O). The animal was placed supine in the angiosurgical suite. ECG, partial pressure of O2, arterial blood pressure, rectal temperature, and end tidal CO2 were continuously monitored and recorded throughout the procedure. An 8F hemostatic introducer was placed in the right femoral artery using a modified Seldinger technique following a cutdown.

Whole blood was taken from the femoral artery for the preparation of fibrin-rich autologous clots. Whole blood (30 mL) was first mixed with citric acid, sodium citrate, and dextrose solution in a 10:1 ratio, followed by the addition of 5.64 mL calcium chloride (CaCl2) solution and 5 mg fibrinogen (F8630, Sigma-Aldrich, St Louis, Michigan, USA). The CaCl2 solution was prepared by dissolving CaCl2 in phosphate buffered solution, resulting in a final concentration of 1.4 g/dL. Clotting was initiated by introducing thrombin (T7513, Sigma Aldrich) into the aforesaid mixture at a concentration of 2.5 NIH units/mL blood. Clots were prepared in silicone tubing of diameters of 1.5, 2.3, and 3.2 mm and allowed to age for approximately 30 min before introduction into the internal carotid artery (ICA).

A 5F or 6F Navien guide catheter was navigated over a microcatheter under X-ray fluoroscopy to the origin of the ICA. All devices were continuously flushed with heparinized saline (3000 U/L). Digital subtraction angiography (DSA) and three-dimensional rotational angiography were acquired to evaluate ICA tortuosity and MCA diameter. A clot diameter slightly larger than the diameter of the MCA was selected. The clot was suspended in contrast in a syringe before injection. With the guide catheter positioned in the origin of the ICA, the clot was introduced via manual injection under fluoroscopic visualization. Injection persisted until the clot entered the MCA. MCA occlusion was confirmed by DSA and the time was recorded. Collateral status was graded and recorded according to a previously developed system.10

MRI and mechanical thrombectomy

Immediately upon occlusion creation and after the clot delivery catheter was removed, the animals were transferred to the adjacent 3-Tesla MRI scanner (Philips, Best, the Netherlands) with the head placed in a 16-channel SENSE phased array knee coil. The MR protocol included time-of-flight vascular imaging (TR/TE 20/4 ms, FA=20o, matrix 332×212), diffusion-weighted imaging (DWI) (TR/TE 2000/76 ms, FA=90o, b-values=0, 1000 s/mm2, NSA=6, matrix 144×144), and dynamic susceptibility contrast (DSC) perfusion imaging (TR/TE 20/4 ms, FA=40o, 60 dynamics, matrix 320×320) following a bolus injection of gadolinium (0.2 mmol/kg, followed by 15 mL saline at 2 mL/s IV) in order to assess areas of perfusion deficit.

Following imaging, the animals were returned to the angiography suite. Prior to MT, DSA confirmed the persistence of an MCA occlusion. A triaxial system was used: 6F shuttle sheath (Cook, Bloomington, Indiana, USA) placed in the common carotid artery, distal access catheter (DAC) 044 (Stryker Neurovascular, Fremont, California, USA) placed in the ICA, and a Headway Duo microcatheter (Microvention, Aliso Viejo, California, USA) navigated over a micro-guidewire (Chikai 014, Asahi Intecc, Tustin, California, USA) distal to the occlusion. Once the clot was crossed with the microcatheter, the guidewire was removed and a small thrombectomy device (either TigerTriever 17, Rapid Medical, Yokneam, Israel11 or Trevo 3mmx20mm, Stryker Neurovascular) was advanced. The thrombectomy device was deployed by unsheathing and the microcatheter was then removed from the system. During the 5 min integration period, a 60 mL Vacu-Lock syringe was attached to the long sheath for aspiration during removal of the stent retriever into the DAC. DSA was acquired after each thrombectomy pass, and subsequent passes were performed until TICI 2b/3 recanalization or a maximum of three passes were reached.

After clot removal, the anesthetized animal was returned to MRI and the imaging protocol was repeated, along with a susceptibility weighted imaging sequence (SWI) (TE/TE 31/7.2 ms, FA 17o, FoV 150 mm, matrix 250×250, slice thickness 1.25 mm, gap −0.625 mm). The animals were then euthanized with an intravenous bolus of sodium pentobarbital (150 mg/kg). Following euthanasia, the brain was extracted for histology. Immediately following extraction, the brain was completely submerged in cold saline and then placed in a −80° freezer for approximately 20 min. A brain matrix was used to partition the brain into 5 mm sections. Each section was stained with 2,3,5-triphenyltetrazolium (TTC) for measurement of the infarct core.12

Image analysis

The DWI-based apparent diffusion coefficient (ADC) maps were thresholded at 0.53×10-3 mm2s-1 to identify infarcted tissue.13 DSC-MRI was used to generate time-to-peak maps, and the perfusion deficit was identified by voxels of the affected hemisphere having a relative delay of 4 s compared with the normal contralateral hemisphere. Failed TTC staining on brain sections was manually outlined to calculate ‘gold standard’ infarct volumes. Tissue volume salvaged by the thrombectomy procedure was the difference between the pre-thrombectomy perfusion deficit and the final ADC lesion volume; furthermore, the percentage of tissue volume salvaged was calculated as the ratio between tissue volume salvaged by the thrombectomy procedure and the initial perfusion deficit volume. The volume of tissue lost between MRI and recanalization was computed by taking the difference between the final and initial ADC lesion volumes. The percentage of initial perfusion deficit tissue lost prior to and during the thrombectomy procedure was calculated as the ratio of volume of tissue lost from first MRI to recanalization to the volume of initial perfusion deficit.


Intra-procedural measurements of the MCA were taken to choose the clot size most appropriate for the vasculature. The average MCA diameter was 1.28 mm (range: 1.17–1.39 mm). In all cases, the 2.3 mm diameter clot was chosen, with standard length of 7 mm. In all animals, occlusion of the M1 segment of the MCA was confirmed angiographically. Four of six animals successfully completed the MT protocol. In two animals, MT was not performed due to clot autolysis leading to spontaneous reperfusion during MRI in one animal and in a second animal the ICA was inaccessible due to vasospasm that was not resolved with vasodilators. A repeated measure t-test of intraoperative heart rate, SpO2, CO2 showed no significant changes between pre- and post-MCA occlusion. In one case, during advancement of the microcatheter prior to placement of the stent retriever, a tear in the ICA was created resulting in the creation of a direct arteriovenous fistula. However, navigation beyond the fistula was achieved, the stent retriever was deployed, and the entire MT procedure was successfully completed. Complete model outcome details are shown in table 1.

Table 1

Subject details

For the four successful thrombectomy models (see representative case in figure 1), reperfusion time was achieved on average 159 min after stroke onset (range 91–204 min). In all four cases, only one pass was required to achieve a successful outcome, either TICI 2b (n=1) or TICI 3 (n=3). Following MT there was no evidence of intracranial perforation or dissection; furthermore, there was no evidence of subarachnoid hemorrhage or intracerebral hematoma on MRI.

Figure 1

(A) Baseline digital subtraction angiogram (DSA) of the right internal carotid artery (arrow: middle cerebral artery, MCA). (B) Occlusion of the right MCA (arrow) after delivery of autologous clot. (C) Thrombectomy device deployed in the right MCA. (D) Final DSA of the right internal carotid artery demonstrating TICI 3 reperfusion after a single thrombectomy pass.

The pre-thrombectomy perfusion deficit area was manually measured and converted to a volume measurement by factoring in the MR slice thickness of 4 mm. The average perfusion deficit area was 14.48 cm3 (range 11.88–16.26 cm3). The pre-thrombectomy ADC lesion volumes were thresholded to obtain an estimate of infarct volume (see representative image in figure 2). Two animals had no evidence of restricted diffusion immediately following MCA occlusion despite large perfusion deficits (16.01 cm3 and 16.26 cm3). The other two animals had post-occlusion restricted diffusion volumes meeting the predefined ADC threshold of 2.70 cm3 and 3.78 cm3, which corresponded to perfusion deficits of 13.77 cm3 and 11.88 cm3, respectively. Following MT, the ADC lesion volumes were 5.2±2.4 cm3, which by linear extrapolation based on brain volume corresponds to 100 cm3 in the human brain.14 15 On average, the thrombectomy procedure salvaged 62.0% (range: 44.15–84.38%) of the tissue initially indicated to be at risk by perfusion imaging. Conversely, the percentage of initial perfusion deficit tissue lost between imaging and reperfusion was 23.7% (range: 15.62–36.24%). All imaging measurements and volume calculations are shown in table 2.

Figure 2

All images are matched coronal sections: (A) Diffusion-weighted imaging (DWI) showing no area restricted diffusion acquired immediately following occlusion of the middle cerebral artery (MCA) and prior to thrombectomy. (B) MRI time-to-peak (TTP) map acquired post-occlusion but prior to thrombectomy showing an area of delayed perfusion of the right MCA territory. (C) MRI TTP map following thrombectomy showing no perfusion delay within the brain. (D) Final DWI showing an MCA territory infarct. (E) Histology showing corresponding lesion.

Table 2

Estimates of ischemic lesion and infarct volumes from MRI

A strong one-to-one correlation between ADC and TTC volumes was found (R2=0.98; figure 3). The average ADC lesion volume was 5.2 cm3 while the average TTC lesion volume was 5.3 cm3.

Figure 3

Final apparent diffusion coefficient (ADC)-thresholded versus histology-derived infarct volume (linear fit R2=0.98). DWI, diffusion-weighted imaging; TTC, 2,3,5-triphenyltetrazolium.


In this study a large animal stroke model that incorporates MT is presented. The development of a stroke model amenable to MT provides a potential paradigm to study neuroprotection strategies and timing of adjunctive thrombolysis coupled with clinical standard of care endovascular treatment. The introduction of MT into the arena of stroke treatment was a revolutionary step in reducing morbidity and mortality of the disease. A significant advance to ensure this treatment is received by the maximum number of patients suffering a large vessel occlusion may be a neuroprotection strategy that provides time for patients to get to comprehensive stroke centers. It is thought that neuroprotective agents have not yet been successful in clinical trials due to the lack of coupling with an effective reperfusion strategy.16 A preclinical model of MT may afford optimization of multiple adjunctive therapies with respect to timing as well as dose and route of administration. The nuances of adjunctive therapy are burdensome for optimization in clinical trials, so we propose an MT model for studying multimodal therapies in a preclinical setting.

There are very well characterized large animal models of focal ischemic stroke in sheep17 18 and pigs19 that deploy a microsurgical approach for either clipping or cauterization of the MCA. However, the presence of the rete mirabile in these species prohibits endovascular access to the anterior circulation for the creation of an embolic MCA occlusion or to perform MT. MT can be simulated in these models by transient occlusion of the MCA; however, specific features of MT such as trauma to the lumen where the device is deployed20 and distal emboli21 are not modeled with this approach. Non-human primate models (reviewed by Mehra et al 8) are theoretically amenable to MT as endovascular access has been documented.22 However, common non-human primate strains (eg, cynomolgus macaques) used in research are hypercoaguable23 and full characterization of embolic stroke is required prior to the addition of MT. The advantages of the canine model include a long history documented in the literature characterizing embolic occlusion of the MCA. Although endovascular access to the intracranial circulation is possible, it is complicated by small and tortuous arteries.

Several important components were integral in the development of this model. It was helpful to pre-screen the animal to choose the left or right MCA for MT based on the extent of ICA tortuosity, the angle between the ICA and the MCA, and MCA diameter. The clot composition included fibrinogen and excluded barium in order to reduce clot hardness and increase elasticity. During the procedure, the use of a triaxial system consisting of a 6F shuttle sheath, a DAC 044 as an intermediate catheter, and a Headway Duo microcatheter allowed for stepwise proximal support in the common carotid artery and ICA, necessary to navigate through the torturous canine ICA.

The introduction of a clot into the MCA of a canine creates a perfusion deficit in the MCA territory; this perfusion deficit indicates critically-hypoperfused tissue that may progress to irreversible infarction without reperfusion.24 25 Furthermore, should reperfusion not be accomplished, the initial ischemic core expands over time.24 26 Clinically, it is established that the goal of MT is to salvage at-risk tissue predicted by the perfusion deficit.27 By assessing the difference between the pre-thrombectomy perfusion deficit and the post-thrombectomy ADC ischemic lesion volume, we showed that MT in a canine model of ischemic stroke is able to mimic this clinical presentation and outcome. The final ADC ischemic lesion never reached the initial perfusion deficit volume, indicating that the intervention was most likely responsible for arresting the progression of infarct evolution. However, a certain percentage of the initial perfusion deficit was lost to infarction prior to performing MT, representing the volume of tissue unable to be saved by MT. This volume of tissue in our model serves as the target of potential neuroprotection strategies needed in the next chapter of stroke treatment.

Some limitations arise in using this model. Vessel tortuosity is a known challenge in the canine cerebrovasculature. Additionally, due to vessel size, the model only has the capacity for small-scale thrombectomy devices and, as such, is not suitable for testing next-generation thrombectomy technology. Balloon-protected thrombectomy has been shown repeatedly to reduce distal emboli28 29 and is associated with improved clinical outcomes.30 A limitation of this model is that the canine ICA is not sufficiently large to accommodate a balloon guide catheter. However, in future experiments it is possible to use balloon guide catheters in the common carotid artery. It is well established in the canine model of MCA occlusion that there is variability of infarct evolution as a function of collateral supply.31–33 Therefore, it is important to use perfusion imaging to determine the extent and severity of the ischemic lesion. Having used six dogs for this experiment with four successful thrombectomies, this study yielded a success rate of 67%. Expanding the number of subjects used in this study would be expected to increase the success rate of accomplishing all aspects of the protocol.


A large animal model of focal ischemic stroke in which MT is performed to recanalize a large vessel occlusion is introduced. This model enables preclinical exploration and optimization of complex multimodal treatment protocols in the setting of clinically relevant MT to advance translational research.



  • Contributors OWB: designed and performed the experiments, analyzed the data, drafted the manuscript, approved the final manuscript. RMK: performed the imaging experiments, analyzed the data, revised the manuscript, approved the final manuscript. J-YC: performed the experiments, revised the manuscript, approved the final manuscript. EN: critical input for optimization of the model, interpreted the data, revised the manuscript, approved the final manuscript. MM: performed the experiments, interpreted the data, revised the manuscript, approved the final manuscript. JC: performed the experiments, analyzed and interpreted the data, approved the final manuscript. ASP: designed the study, interpreted the data, revised the manuscript, approved the final manuscript. MJG: designed the study, designed and performed the experiments, analyzed and interpreted the data, drafted the manuscript, approved the final manuscript and agrees to be accountable for the accuracy and integrity of the work.

  • Funding Partially supported by research funding from Rapid Medical. JC was supported by research grants from the Fulbright Program, the Philippe Foundation, and the French Society of Radiology (SFR-CERF). The content is solely the responsibility of the authors and does not reflect the opinions of any sponsors.

  • Competing interests OWB and RMK declare that they have no competing interests. EN: Fee-for-service consulting for Rapid Medical. MM, J-YC: Fee-for-service consulting for Stryker Neurovascular and InNeuroCo. JC: has received educational scholarships from Medtronic Neurovascular and Microvention/Terumo. ASP: consultant for Medtronic Neurovascular and Stryker Neurovascular; research grants from Medtronic Neurovascular and Stryker Neurovascular. MJG: has been a consultant on a fee-per-hour basis for Cerenovus, Imperative Care, Mivi Neurosciences, Phenox, Route 92 Medical, Stryker Neurovascular; holds stock in Imperative Care and Neurogami; and has received research support from the National Institutes of Health (NIH), the United States–Israel Binational Science Foundation, Anaconda, Cerenovus, Cook Medical, Gentuity, Imperative Care, InNeuroCo, Magneto, Microvention, Medtronic Neurovascular, MIVI Neurosciences, Neuravi, Neurogami, Philips Healthcare, Rapid Medical, Route 92 Medical, Stryker Neurovascular, Syntheon, and the Wyss Institute.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data presented in the manuscript.

  • Presented at This work has been presented in part at the Society of NeuroInterventional Surgery 15th Annual Meeting (SNIS 2018).

  • Patient consent for publication Not required.