Robotics for Revision Spine Surgery

Introduction

Despite advances in our understanding of spinal pathologies, diagnostics, implant designs, and biological adjuncts, there is a significantly sized cohort of patients who fail to achieve resolution of pain following surgery or have recurrent symptoms after surgery. Revision spine surgery tends to have diminished odds for success that continues to decline with each additional surgery. These patients also present technical challenges for the spine surgeon with regards to instrumentation when anatomical landmarks are lost secondary to fusion masses and scar tissue.

The placement of spinal instrumentation, especially pedicle screws, is an important step in modern spine surgery. Pedicle screw placement is a technique that requires fine manipulation of trajectories to deeply seated bony structures that are accessed through a small corridor. No task is better suited for the use of robotic assistance than placement of pedicle screw instrumentation, as robots are indefatigable and are able to perform repetitive tasks with precision and reproducible outcomes. In this regard, robotic assistance with spinal instrumentation provides a means of delaying surgeon fatigue and allowing focus and energy to be diverted to the task of working around the spinal cord and nerve roots.

Robotic-assisted surgery for simple and complex procedures is routinely performed in other surgical subspecialties such as cardiac and urological surgery. Image-guided robotics is relatively novel in the field of spine surgery but its integration into current practice has been shown to be safe and allows surgeons to visualize and navigate complex revision spine cases. In this chapter, we will discuss the role of robotics in revision spine surgery.

Robotics as an Emerging Technology

As robotic technology continues to advance, its integration into medicine can be seen across various specialties. It has been used to improve outcomes for many surgical procedures, including in the spine, where robotic assistance has seen its greatest impact in pedicle screw placement. Originally, pedicle screw placement was described using fluoroscopic guidance. As surgeons developed a better understanding of the spinal anatomy, Kim et al. described a freehand pedicle screw placement technique using anatomic landmarks. This allowed for faster screw placement while drastically reducing the amount of radiation exposure for both the patient and surgeon. The latest iteration of pedicle screw placement techniques has been based on navigation using fluoroscopy and computed tomography (CT) scans, and robotic technology has been developed to increase accuracy during pedicle screw placement. The use of navigation allows for screw placement without relying on anatomy, which becomes critical in revision or tumor cases, where the native anatomy can be significantly altered, or with minimally invasive cases, where exposure and visualization may be limited. Studies looking at the efficacy of robotic guidance in spine surgery are relatively new. Only a few clinical studies have been published over the past decade that have compared freehand to robot-assisted pedicle screw placements.

A few cadaveric studies have looked at robotic screw placement relative to freehand placement. Lieberman et al. looked at 234 pedicle screws in thoracolumbar cadaveric models. Using the Rampersaud classification on postoperative CT scan, they found 99% satisfactory screw placement in the robotic group (197 screws) versus 95% satisfactory screw placement in the freehand group (37 screws). There were no significant differences in procedure duration. In 2015, Fujishiro et al. used cadaveric models to determine if preoperative screw plan correlated with actual implementation using the robot. They placed wires in 216 pedicles using starting points determined by the robot. No significant differences were seen in starting point or in trajectory (at a depth of 30 mm) between the preoperative plan or the wire, in either the axial or sagittal planes. This study helped confirm that the robot not only reliably places the start point as planned, but also can set the trajectory predictably at a depth of 30 mm (past the length of the pedicle).

In the clinical setting, most published studies on robots have been retrospective reviews; however, Ringel et al. performed a prospective randomized controlled trial in 2012 in 60 patients. They placed 298 screws: 152 using freehand technique and 146 using robotic guidance. They found 93% of freehand screws were in good position (grade A or B using the Gertzbein-Robbins scale), whereas 85% were in good position using robotic guidance. Surgical time was statistically significantly shorter in the freehand group (85 minutes) versus the robotic-guidance group (95 minutes). Similarly, Kantelhardt et al. performed a retrospective study in 2011 in 112 patients with six different surgeons. They found 91.4% of screws were accurately placed in the free hand group, versus 94.5% in the robotic-guided group. Within the robotic-guided group, they also compared a traditional open midline approach (20 patients) with a percutaneous minimally invasive approach (35 patients) and found no significant difference in screw accuracy. There was no difference in procedure time for robotic guidance versus freehand placement. However, there was significantly less radiation exposure (based on total x-ray exposure time) in the robotic-guided group. Along the same lines, Schatlo et al. looked at 95 patients with degenerative lumbar spine disease and compared 244 screws placed with robotic assistance with 163 screws placed with fluoroscopic guidance. Using the Gertzbein-Robbins scale, they found that 83.6% had a perfect trajectory (A) and 91.4% were graded either A or B in the robot group. Similarly, 79.8% had a perfect trajectory (A) and 87.2% were clinically acceptable (grade A or B) in the fluoroscopy group. There was no significant difference between the robotic and fluoroscopic cohorts with regard to clinically acceptable screw placement. Within the robotic-guidance group, Schatlo et al. similarly found no difference in accuracy using a traditional open approach versus a minimally invasive approach. A larger study by Molliqaj et al. in 2017 looked at 880 screws placed in 169 patients with varying pathologies (degenerative disease, tumor, trauma) with either the freehand technique or with robotic guidance, and also found no difference in accuracy (using the Gertzbein-Robbins scale). Of the screws, 93.4% were “clinically acceptable” in the robotic-guidance group versus the freehand group (88.9%); this was a statistically significant difference. Solomiichuk et al. looked at 406 screws placed in tumor cases and found no significant difference in accuracy between the two using the Gertzbein-Robbins scale. Keric et al. looked at 1857 screws placed using a minimally invasive approach between 2011 and 2016 and found that 96.9% had acceptable placement using the Weisner scale. Of note, 7 cases (of 413) were converted to a traditional open approach owing to registration errors.

Regarding radiographic parameters, Kuo et al. published a study looking at intraoperative accuracy of robotic guidance by assessing the placement of the wire (before screw placement) relative to the preoperatively planned screw trajectory by using secondary registration intraoperatively. A total of 317 screws were placed in 64 patients, with 98.7% of wires within 3 mm of the planned trajectory. The authors noted the general trend was for the wire to be caudal and lateral, a common pattern seen with robotic screw placement. A study by van Djik et al. looked retrospectively at 112 patients with 487 screws placed using a minimally invasive approach and found the mean deviation of the entry point to be 2.0±1.2 mm, and mean angle of insertion was 2.2±1.8 degrees and 2.9±2.4 degrees. Ultimately, 97.8% of screws were safely placed (<2 mm) using the Gertzbein-Robbins scale. None of the screws required revision surgery. Laudato et al. compared screws placed using freehand, navigation, and robotic-guidance techniques. They found no statistical difference between the three groups using the Rampersaud criteria: 6.7% of screws were considered malpositioned in the freehand group, 4.2% in the navigation group, and 4.7% in the robotic group. Regarding S2 alar-iliac (S2AI) screw placement, which can be technically challenging using the freehand technique, a study by Shillingford et al. found that there was no difference in overall accuracy between the two techniques. Looking at 105 total S2AI screws, they found that both the horizontal angle (in the axial plane) and the S2AI to S1 screw angle were not significantly different between the two groups. No intraoperative neurovascular complications were noted for either group.

These studies illustrate that robotic screw placement is safe and accurate, similar to previous fluoroscopic and freehand techniques. The cadaveric study by Fujishiro et al. showed that screw placement does in fact match the planned preoperative template, allowing surgeons to make specific operative plans and maximize screw length and width before the case. The various clinical studies further demonstrated that accuracy of robotic screw placement was similar to both freehand and traditional fluoroscopic techniques, as confirmed by postoperative CT imaging. There was also no significant difference in adverse outcomes or complications. There were differing conclusions on whether the robot added operative time. Although there was a statistical difference in one study, clinically this may be a negligible difference. Operative time also depends heavily on the surgeon’s comfort with the technology, as well as the staff and assistants. Lastly, one study also showed the robot decreased intraoperative radiation exposure relative to the freehand technique.

Technique

A preoperative CT scan is performed and a specific protocol must be implemented. However, no additional special equipment or expertise is required of the CT facility. Next, preoperative planning is performed by the surgeon using the corresponding software, where each screw size and trajectory is mapped on the CT scan. Once planning is complete, the plan may be uploaded to the workstation before the case begins ( Figs. 17.1 and 17.2 ).

Once in the operating room, the patient is positioned prone on the operating room table in the standard fashion. At this time, the robot may be mounted to the bed. Of note, it is easiest to mount the robot if the bed is parallel to the ground, after which the bed may be adjusted as necessary (reverse-Trendelenburg, etc.). Once the robot is mounted, the patient is prepped and draped in standard fashion ( Fig. 17.3 ). An additional drape is placed over the robot, as well as additional half-sheets around the base of the robot, to maintain sterility. Once draped, the arms may be rotated out of the field until exposure is complete and screws are ready to be placed. Of note, exposure will not necessarily be as extensive as traditional freehand technique as only the starting points need to be exposed, without requiring exposure of surrounding landmarks.