• intervention

This issue of the Journal of NeuroInterventional Surgery (JNIS) features the first three reports of robotic assistance in neuroendovascular procedures. Pereira and colleagues present the first-in-man, robotic-assisted, intracranial aneurysm treatment.1 Nogueira and co-authors report the first four human patients treated with robotic-assisted carotid stenting.2 Finally, Sajja et al describe their experience in diagnostic angiography, including transradial approaches, and carotid stenting in a series of 10 patients.3

In concept, this nascent technology holds tremendous promise for our field. The ability to use robots to deliver remote, interventional stroke care in rural and other less densely populated areas is a potential solution to one of the greatest challenges facing emergent large vessel occlusion (ELVO) patients. If successfully implemented, this technology could improve patient outcomes and revolutionize stroke care worldwide.

In the three reports published this month, operators performed robotic-assisted, neurodiagnostic and interventional procedures with the robotic console located just outside of the neuroangiography suite. All three studies employed the CorPath GFX system (Corindus Inc, Waltham, MA). The authors point out that these feasibility demonstrations are the first iterative steps toward this ultimate vision of remote stroke care. For locally performed neurointerventional procedures, the primary potential advantage of robotic assistance for the patient relates to the ability of the operator to perform precise device manipulations that are beyond the level of human dexterity. Additional potential benefits for the operator include reduced radiation exposure and reduced orthopedic injury related to time spent standing in lead aprons.

Despite the enormous future promise of remote operation and the theoretical existing benefits for local use, many important questions remain unanswered. To be routinely implemented in neurointervention going forward, the major issues of regulatory clearance, device optimization, operator training and, most critically, the generation of high quality evidence from prospective clinical trials confirming the relative safety and efficacy of these procedures are of obvious and paramount importance.

The CorPath GFX system was developed and optimized for interventional procedures using in-line devices, specifically, angioplasty balloons and balloon-mounted stents. The system currently only has US Food and Drug Administration (FDA) clearance for coronary and peripheral interventions. Since, from the standpoint of the FDA, carotid interventions are considered “peripheral” (non-neurological) procedures, the cases in the Nogueira report were performed under this FDA clearance within the context of routine clinical practice. Sajja and co-authors argue that since neuroangiography involves essentially the same catheter placement as carotid stenting, these procedures were also considered within the spectrum of “peripheral vascular procedures” and, hence, also on-label.

In Canada, the CorPath GFX system does not currently have regulatory clearance. Pereira and co-authors obtained approval for use of the device through a Special Access application to Health Canada. This Special Access Program (similar to Compassionate Use in the USA) is a unique mechanism available to practicing Canadian physicians. Appropriate providers may obtain permission from Health Canada to use unapproved devices to treat potentially life-threatening or serious diseases. This exemption is specifically designed for pathologies that are not amenable to standard therapies. The authors argue that their extensive experience using the CorPath GFX system in animal and patient-specific simulation experiments validated their assessment that the robot provided enhanced precision in stent deployment and reliable efficacy in coil embolization.

From the descriptions in these reports, it is clear that many aspects of the current robotic system have not yet been optimized for neurovascular applications. Substantial portions of the procedures—for example, the critical steps of carotid stent deployment and catheterization of a bovine aortic arch—often could not be performed robotically and required conversion to manual techniques. These issues raise concerns as to the suitability of using this technology to perform routine procedures in its current stage of evolution.

Given the novelty of this technology and its first use in human beings, the obligate inexperience of the operators certainly comes with added risk—even under anatomically optimal conditions. One could argue that a major disadvantage of robotic assistance is the loss of the haptic feedback critical to the safe performance of procedures such as coil embolization and carotid stenting. The experienced operator recognizes when pushing too hard to deliver devices to a targeted landing zone or when devices are not moving in one-to-one synchronicity. It is challenging to justify how performing these technically straightforward steps with robotic assistance is advantageous, especially if such maneuvers take additional time, introduce complexity, and rely on technology that is not yet developed for neurointerventional use. The authors of these studies argue that the disadvantages of the loss of this haptic feedback and the conversion of a highly familiar user interface (microwires and microcatheters) to a novel user interface (joystick controls and touchscreens) are mitigated by having better control over micro-movements during stent placement, coil embolization and diagnostic angiography.

When considering the larger picture of remote robotic interventions, the significance of these cases and the motivation behind demonstrating feasibility are clear. Some concerns such as perceived complexity, handling joysticks instead of catheters, and additional procedure time may in fact be addressed with sufficient simulator training and greater experience. The vision of a multitude of robots stationed in rural hospitals or throughout satellites of a hospital network that could be operated by a single team of expert neurointerventionalists at a central location holds tremendous potential. Undoubtedly, this vision is intriguing and forward looking, but it is also not the current reality. The central question is how to get from here to there while not compromising patient safety. The feasibility of emergent robotic thrombectomies will depend first on whether robotic neurointerventions can be reliably performed in controlled, elective procedures with measurable safety and efficacy.

Neurointervention is a rapidly moving, technologically driven field where innovation is constant and paradigm changes are the norm. While we often bemoan the enormous burden placed on us by regulatory bodies and ethics committees, it seems somewhat unsettling that robotic systems could be so rapidly cleared for routine clinical use and applied to supplant established, manual neurointerventional procedures in the absence of evidence to support at least equivalent safety and efficacy. Ideally, future cases should be performed within the context of a prospective, externally monitored research study. It is intuitive that robotics may have an important role to play in neurointervention, but it is also critical that we proceed in this direction in a careful, systematic and impartial manner.