21.2.1 Pedicle Screw Accuracy

Early understanding of the accuracy of pedicle screw placement by the Mazor robot was described by Sukovich et al who presented a retrospective review of 98 screws in 14 patients and found that 96% of the screws were within 1 to 2 mm of their planned trajectory. Furthermore, they observed no pedicle breaches from screw malposition. ¹

To more systematically categorize pedicle screw positioning, many of the later studies utilized the CT-based Gertzbein and Robbins classification system (GRS). ² In this system, the screw’s placement is given a grade of A through E. Grade A is no breach; a breach of less than 2 mm is Grade B; a breach of 2 to less than 4 mm is Grade C; a breach of 4 to less than 6 mm is Grade D; a breach of greater than 6 mm is Grade E. To help guide surgeons’ decision making, the authors made the determination that Grades A and B were qualified as acceptable placement.

In a retrospective study utilizing the GRS system, Onen et al studied a series of 27 patients in whom 136 robotically assisted screws were placed. ³ They noted 91.2% of screws were deemed Grade A and 7.4% of screws were Grade B. Furthermore, Schatlo et al and van Dijk et al conducted similar retrospective reviews on the placement of 1,265 and 494 robotically assisted screws, respectively. Both studies demonstrated successful placement (GRS Grade A or B) of over 96% of screws. ^{4 ,​ 5} The Mazor system was also studied in prospective fashion using the GRS classification for its use in pedicle screw placement in patients undergoing posterior lumbar interbody fusion (PLIF). ⁶ After a review of 122 screws (31 patients), they determined that over 99% were GRS Grade A or B.

The largest retrospective review to date was performed by Devito et al. They had planned for 3,912 screws to be placed in 682 patients. However, 3,271 screws (83.6%) were placed with full robotic guidance, while the remaining were started with the robot but finished manually. ⁷ The initial fluoroscopic evaluation determined that 98% of the screws were in an acceptable position. For the 646 screws in patients who had a subsequent CT scan, over 98.3% were designated as either GRS Grade A or B. Of the remaining screws rated Grade C or lower, 1.4% breached between 2 and 4 mm and only 2 screws (0.3%) violated the pedicle wall by more than 4 mm. Neurologic symptoms were found in four patients; however, following revision surgery, no permanent deficits were noted.

In an effort to help optimize usage of the Mazor robot, Kuo et al developed an intraoperative assessment of robotic pedicle screw placement through a secondary registration. ⁸ The authors used the Mazor robot to place K-wires and their position was subsequently rechecked with biplanar fluoroscopy. This group determined that secondary registration increased the accuracy rate of robotic pedicle screw placement. ⁸ There were 317 K-wires placed and only 6% (19 wires) were malpositioned by more than 3 mm. Upon repositioning, 15 of 19 were improved. The final four wires required manual placement for an ultimate accuracy rate of 98.7%.

21.2.2 Comparative Studies

In addition to evaluating the accuracy of the Mazor system, several groups have sought to help stratify the utility of robotic guidance by comparing it to a traditional freehand pedicle screw placement and/or CAN.

Perhaps the most well-performed study on the topic was a single-center, prospective, randomized controlled study by Ringel and colleagues. Interestingly, this study produced the only results, to our knowledge, that demonstrate inferiority of pedicle screw placement with the SpineAssist/Mazor robot as compared to the freehand technique. ⁹ In this study, 60 patients were randomly assigned into a percutaneous robotically assisted instrumentation cohort or an open, freehand technique cohort. The robotically assisted cohort demonstrated acceptably placed screws in 85% of those attempted (146 screws). The freehand cohort showed that 93% of screws were acceptably placed. Furthermore, they noted that the freehand technique showed shorter surgical time (84 vs. 95 minutes). The group also found that when the robotically assisted screws were malpositioned, they tended to be placed laterally. The authors determined that the robotic positioning was vulnerable to lateral deviation due to the fact that there was slippage or skidding of the drill sleeves off the lateral aspect of the facet.

In another randomized controlled trial, Hyun et al randomized 60 patients to receive either robotic or freehand pedicle screws. 100% accuracy was demonstrated in a robotically assisted group as compared to 98.6% in the freehand group. ¹⁰ They noted that in the freehand group, there was a violation of one proximal facet, while there were no such instances in the robotic cohort.

A third randomized controlled trial was performed by Kim et al who compared pedicle screw placement using robot and freehand techniques in patients undergoing minimally invasive PLIF. ¹¹ The team demonstrated a 99.4% accuracy rate in both groups; however, they noted proximal joint violations in 15.9% of freehand screws (13 screws) but no joint violations in the robotically placed screws.

Kantelhardt et al demonstrated 94.5% accuracy with the SpineAssist/Mazor model and 91.5% accuracy with a freehand, fluoroscopically assisted technique in a nonrandomized retrospective study. ¹² They also noted that those who underwent robotic-assisted surgery required less opioids, had shorter hospitalization, and lower rates of adverse events as compared to conventional screw placement. Furthermore, these benefits were exaggerated in patients who underwent percutaneous robot-assisted procedures.

Schizas and colleagues, in a prospective study, reported 95% accuracy for robot-assisted pedicle screw placement (11 patients, 64 screws) as compared to 92% with fluoroscopically guided insertion (23 patients, 64 screws). ¹³ Another study, as part of a review on robotic spine surgery, described a preliminary prospectively collected data set that demonstrated a 99% accuracy rate of robotic lumbosacral pedicle screw placement compared to 98% accuracy with a fluoroscopically guided technique and a 92% success rate with CAN. ¹⁴

A retrospective, case-matched study by Schatlo et al found a nonsignificant difference in the proportion of acceptable pedicle screws placed between the Mazor system (91.4%) and conventional freehand technique (87.1%). ¹⁵ Notably, in their series, the group had one freehand screw that required a revision procedure due to an iatrogenically induced radiculopathy. Another retrospective cohort study of patients with spondylodiskitis by Keric et al demonstrated superiority in the accuracy of robotically assisted pedicle screw and that they were less likely to require revision for malposition or loosening. ¹⁶

Roser et al performed a randomized controlled trial with three groups: traditional freehand technique, standard neuronavigation, and robotically assisted pedicle screw placement. ¹⁴ In this study, an “accurate” screw was defined as GRS Group A, which differed from most of the studies in the literature that denoted both GRS Groups A and B to be considered accurate. The Mazor robot (99% accuracy) was superior to both the freehand technique (97.5% accuracy) and the neuronavigation technique (92% accuracy). When GRS Group B screws were included in what was considered “acceptable,” the accuracy rates increased to 99% with the robot, 100% with the freehand technique, and 97.2% with neuronavigation.