OBJECTIVERobot-guided pedicle screw placement is an established technique for the placement of pedicle screws. However, most studies have focused on degenerative disease. In this paper, the authors focus on metastatic spinal disease, which is associated with osteolysis. The associated lack of dense bone may potentially affect the automatic recognition accuracy of radiography-based surgical assistance systems. The aim of the present study is to compare the accuracy of the SpineAssist robot system with conventional fluoroscopy-guided pedicle screw placement for thoracolumbar metastatic spinal disease.METHODSSeventy patients with metastatic spinal disease who required instrumentation were included in this retrospective matched-cohort study. All 70 patients underwent surgery performed by the same team of experienced surgeons. The decision to use robot-assisted or fluoroscopy-guided pedicle screw placement was based the availability of the robot system. In patients who underwent surgery with robot guidance, pedicle screws were inserted after preoperative planning and intraoperative fluoroscopic matching. In the “conventional” group, anatomical landmarks and anteroposterior and lateral fluoroscopy guided placement of the pedicle screws. The primary outcome measure was the accuracy of screw placement on the Gertzbein-Robbins scale. Grades A and B (< 2-mm pedicle breach) were considered clinically acceptable, and all other grades indicated misplacement. Secondary outcome measures included an intergroup comparison of direction of screw misplacement, surgical site infection, and radiation exposure.RESULTSA total of 406 screws were placed at 206 levels. Sixty-one (29.6%) surgically treated levels were in the upper thoracic spine (T1–6), 74 (35.9%) were in the lower thoracic spine, and the remaining 71 (34.4%) were in the lumbosacral region. In the robot-assisted group (Group I; n = 35, 192 screws), trajectories were Grade A or B in 162 (84.4%) of screws. The misplacement rate was 15.6% (30 of 192 screws). In the conventional group (Group II; n = 35, 214 screws), 83.6% (179 of 214) of screw trajectories were acceptable, with a misplacement rate of 16.4% (35 of 214). There was no difference in screw accuracy between the groups (chi-square, 2-tailed Fisher’s exact, p = 0.89). One screw misplacement in the fluoroscopy group required a second surgery (0.5%), but no revisions were required in the robot group. There was no difference in surgical site infections between the 2 groups (Group I, 5 patients [14.3%]; Group II, 8 patients [22.9%]) or in the duration of surgery between the 2 groups (Group I, 226.1 ± 78.8 minutes; Group II, 264.1 ± 124.3 minutes; p = 0.13). There was also no difference in radiation time between the groups (Group I, 138.2 ± 73.0 seconds; Group II, 126.5 ± 95.6 seconds; p = 0.61), but the radiation intensity was higher in the robot group (Group I, 2.8 ± 0.2 mAs; Group II, 2.0 ± 0.6 mAs; p < 0.01).CONCLUSIONSPedicle screw placement for metastatic disease in the thoracolumbar spine can be performed effectively and safely using robot-guided assistance. Based on this retrospective analysis, accuracy, radiation time, and postoperative infection rates are comparable to those of the conventional technique.