Robotic Stroke Thrombectomy: A Feasibility and Efficacy Study in Flow Models

Article information

J Stroke. 2025;27(2):266-269

Publication date (electronic) : 2025 May 31

doi : https://doi.org/10.5853/jos.2024.05057

Cameron J. Williams^,1,2,3,*, Hal Rice^4,*, Bernard Yan^1,2,3, Laetitia de Villiers⁴, Vinicius Carraro do Nascimento⁴, Peter J. Mitchell^2,5, Nathan W. Manning⁶, Leonid Churilov³, Mark W. Parsons⁷, Stephen M. Davis^1,3,†, Geoffrey A. Donnan^1,3,†

¹Melbourne Brain Centre, Department of Neurology, Royal Melbourne Hospital, Parkville, VIC, Australia

²Department of Radiology, Royal Melbourne Hospital, Parkville, VIC, Australia

³Department of Medicine, University of Melbourne, Parkville, VIC, Australia

⁴Department of Interventional Neuroradiology, Gold Coast University Hospital, Southport, QLD, Australia

⁵Department of Radiology, University of Melbourne, Parkville, VIC, Australia

⁶Department of Neurointervention, Liverpool Hospital, Liverpool, NSW, Australia

⁷Department of Neurology, Liverpool Hospital, Liverpool, NSW, Australia

Correspondence: Cameron J. Williams Melbourne Brain Centre, Department of Neurology, Royal Melbourne Hospital, 300 Grattan Street, Parkville, Victoria, Australia, 3050 Tel: +61-3-9342-8448 E-mail: cameron.williams8@mh.org.au

*These authors contributed equally as first author.

†These authors contributed equally as last author.

Received 2024 November 21; Accepted 2025 January 15.

Dear Sir:

Endovascular thrombectomy (EVT) is the most effective treatment for acute ischemic stroke (AIS) with large vessel occlusion (LVO). The most important determinant for good patient outcomes is the time from symptom onset to reperfusion [1].

In the Western world, approximately 33%–40% of the population resides in rural and remote areas, often relying on prolonged aeromedical interhospital transfers to metropolitan centers where EVT is provided [2]. This results in delayed EVT, which can often lead to poorer clinical outcomes.

Economic analysis and modeling suggest that remote robotic EVT in AIS would be a cost-effective, innovative strategy and could potentially extend care to underserved communities and rural areas [3]. Despite several publications confirming the feasibility, efficacy, and safety of telerobotics in percutaneous coronary intervention (PCI) [4], cerebral aneurysms [5], and carotid stenting [6], there is no literature describing telerobotic EVT in LVO patients.

Our study aimed to test the hypothesis that robotic EVT performed in an in vitro model would provide evidence to establish its feasibility, efficacy, and safety. Our vision is to build on this pre-clinical research as a precursor to long-distance remote EVT in these same models, prior to our ultimate goal to perform remote EVT in stroke patients. This could revolutionize access to the most effective modern stroke therapy for patients in rural and remote regions.

This study was a prospective, single-institution, open-label, nonrandomized, blinded outcome study, approved by the Human Research Ethics Committee (HREC; approval number MDF/2022/MDF/89654). We manually placed synthetic thrombus (Thrombotech, Life Model Designs, https://lmd3d.com) to replicate an LVO in three-dimensional (3D) printed models (Life Model Designs, https://lm3d.com) of cerebral vasculature. Robotic EVT was performed using the CorPath GRX system (Corindus, Siemens Healthineers Company, Waltham, MA, USA), which is approved by the Australian Government Therapeutic Goods Administration (TGA) for neurointerventional procedures in Australia (Australian Register of Therapeutic Goods [ARTG] ID 900971). Three qualified neurointerventionalists with varying levels of robotic experience performed robotic EVT as the primary proceduralists while positioned in a separate control room. Navigation of a microcatheter and microwire through the occluded vessel segment was undertaken by the proceduralist with subsequent deployment of a stentriever before retraction of the deployed stentriever into the guide catheter without direct aspiration. Arterial access and guide catheter placement were not assessed.

The primary outcome was successful recanalization with removal of the thrombus within the model, resulting in an arterial occlusive lesion (AOL) score of 2–3. The AOL recanalization score was selected rather than the modified Thrombolysis in Cerebral Infarction scale, given the rudimentary vasculature of the in vitro model.

Successful robotic navigation and deployment of the devices (microcatheters, microwires, and stentrievers), non-target embolization, total fluoroscopy time, total procedure time, and contrast volume were also assessed.

Safety outcomes included rupture of the model during robotic EVT and conversion to standard manual EVT due to machine and/or network malfunction.

Due to the descriptive nature of the study, outcome data were summarized using medians and interquartile ranges (IQRs) for continuous data and numbers (percentages) for categorical data.

Between January 2023 and August 2023, 27 robotic EVT procedures were performed in 3D-printed flow models (Supplementary Figure 1) with LVO at a single cerebrovascular robotic center. All cases were performed using a stentriever without concurrent aspiration. Each neurointerventionist performed nine procedures as the primary proceduralist. The characteristics of the flow models are summarized in Table 1.

Table 1.

Characteristics of flow models resembling large vessel occlusion for robotic endovascular thrombectomy

For robotic EVT, 100% of cases had AOL 3 (complete recanalization). Similarly, all had successful navigation and deployment of devices (microcatheters, microwires, and stentrievers) within the intracranial segments of the flow models. There were no cases of non-target embolization, model perforation, or machine failure requiring manual conversion. The secondary outcomes are further summarized in Table 2.

Table 2.

Primary, secondary, and safety outcomes of robotic EVT in flow models

Our study demonstrated that robotic EVT had a 100% success rate in achieving complete recanalization without perforation and without non-target embolization, using in vitro models.

Despite small numbers, our study was shown to have the highest procedural success rate compared to other published robotic EVT research using models [7]. We also showed a trend toward shorter procedural times with increased operator experience, consistent with previous robotic-assisted diagnostic cerebral angiography and carotid artery stenting studies [8]. Although the minimum number of preclinical robotic interventions required prior to first-in-human EVT studies is unknown, a level of operator “prowess” and robotic procedural competency was achieved after as few as 10 procedures preceding robotic PCI in a clinical setting [9].

The total procedural time to achieve successful recanalization with robotic EVT in our study was 15 minutes (IQR 12–19 minutes). Allowing for additional procedural time for arterial puncture and placement of the guide catheter in the access artery in a clinical setting, this time metric appears comparable to the median puncture-to-recanalization time of 43 minutes in the EXTEND-IA trial. We believe that any potential for initially longer procedural times with robotic EVT in a rural setting will likely be mitigated by the significant time saved avoiding prolonged interhospital transfers with current practice.

In our study, there were no cases of machine failure or manual conversion. Perhaps this was because our study was performed using wired connection in a local-remote setting; however, no significant differences in procedural or clinical outcomes between wired and wireless networks have been demonstrated with remote PCI over long distances [10].

The translation from in vitro success to human application is complex. Larger randomized studies with robotic platforms specifically designed for neurointervention are required to elaborate on our findings, with the aim to increase generalizability to a broader population. Similarly, clarity around the medico-legal framework and management of unforeseen intraoperative complications in a remote setting will be required.

This study demonstrates that robotic EVT is potentially feasible and effective when performed in flow models. Further research is required before its potential future application in first-in-human studies, thus following the step-wise approach used in the development of remote PCI. This innovative technology has the potential to provide earlier recanalization and, hence, improved clinical outcomes in LVO patients living outside metropolitan regions requiring EVT.