Accuracy of robot-guided versus freehand fluoroscopy-assisted pedicle screw insertion in thoracolumbar spinal surgery

OBJECTIVE The quest to improve the safety and accuracy and decrease the invasiveness of pedicle screw placement in spine surgery has led to a markedly increased interest in robotic technology. The SpineAssist from Mazor is one of the most widely distributed robotic systems. The aim of this study was to compare the accuracy of robot-guided and conventional freehand fluoroscopy-guided pedicle screw placement in thoracolumbar surgery. METHODS This study is a retrospective series of 169 patients (83 women [49%]) who underwent placement of pedicle screw instrumentation from 2007 to 2015 in 2 reference centers. Pathological entities included degenerative disorders, tumors, and traumatic cases. In the robot-assisted cohort (98 patients, 439 screws), pedicle screws were inserted with robotic assistance. In the freehand fluoroscopy-guided cohort (71 patients, 441 screws), screws were inserted using anatomical landmarks and lateral fluoroscopic guidance. Patients treated before 2009 were included in the fluoroscopy cohort, whereas those treated since mid-2009 (when the robot was acquired) were included in the robot cohort. Since then, the decision to operate using robotic assistance or conventional freehand technique has been based on surgeon preference and logistics. The accuracy of screw placement was assessed based on the Gertzbein-Robbins scale by a neuroradiologist blinded to treatment group. The radiological slice with the largest visible deviation from the pedicle was chosen for grading. A pedicle breach of 2 mm or less was deemed acceptable (Grades A and B) while deviations greater than 2 mm (Grades C, D, and E) were classified as misplacements. RESULTS In the robot-assisted cohort, a perfect trajectory (Grade A) was observed for 366 screws (83.4%). The remaining screws were Grades B (n = 44 [10%]), C (n = 15 [3.4%]), D (n = 8 [1.8%]), and E (n = 6 [1.4%]). In the fluoroscopy-guided group, a completely intrapedicular course graded as A was found in 76% (n = 335). The remaining screws were Grades B (n = 57 [12.9%]), C (n = 29 [6.6%]), D (n = 12 [2.7%]), and E (n = 8 [1.8%]). The proportion of non-misplaced screws (corresponding to Gertzbein-Robbins Grades A and B) was higher in the robot-assisted group (93.4%) than the freehand fluoroscopy group (88.9%) (p = 0.005). CONCLUSIONS The authors' retrospective case review found that robot-guided pedicle screw placement is a safe, useful, and potentially more accurate alternative to the conventional freehand technique for the placement of thoracolumbar spinal instrumentation.