Accuracy of imaging grading in comparison to open laminectomy to evaluate pedicle screws positioning.

Study design: Prospective experimental study. Objective: To compare the accuracy of O-Arm-acquired radiographic and computed tomography (CT) evaluation of thoracic pedicle screw placement with open laminectomy in a simulation laboratory. Summary of background data: Improving surgical safety and procedural efficiency during thoracic posterior spine instrumentation is essential for decreasing complication rates and possible related risks. The most common way of verifying the position of pedicle screws during the surgical procedure and immediately postoperatively is to acquire intraoperative fluoroscopic images and plain radiographs of the spine, respectively. Laboratory simulated surgery is a valuable tool to evaluate the accuracy of those exams. Methods: Twenty simulation models of scoliosis from T3 to T7 were instrumented by five spine fellows (total of 200 pedicle screws), followed by radiographic and CT images acquired with the assistance of the O-Arm which were evaluated by three independent raters. A fellowship-trained spine neurosurgeon performed laminectomies on the instrumented levels and assessed pedicle integrity (gold standard). Results: Forty-eight breaches were identified in the axial direct view after laminectomy. Of those, eighteen breaches were classified as unacceptable. Regarding the sagittal direct view, four breaches were observed, three of which were classified as unacceptable. Overall, both O-arm radiographic and CT evaluations had a significantly high negative predicted value but a low positive predicted value to identify unacceptable breaches, especially in the sagittal plane. The frequency of missed breaches by all three examiners was high, particularly in the sagittal plane. Conclusion: Postoperative evaluation of pedicle screws using O-arm-acquired radiographic or CT images may underdiagnose the presence of breaches. In our study, sagittal breaches were more difficult to diagnose than axial breaches. Although most breaches do not have clinical repercussions, this study suggests that this modality of postoperative radiographic assessment may be inaccurate. Level of evidence: 4.