Safe Pedicle Screw Placement in Thoracic Scoliotic Curves Using t-EMG

Abstract

A cross-sectional study of nonconsecutive cases (level III evidence). In a series of young patients with thoracic scoliosis who were treated with pedicle screw constructs, data obtained from triggered electromyography (t-EMG) screw stimulation and postoperative computed tomographic scans were matched to find different threshold limits for the safe placement of pedicle screws at the concavity (CC) and convexity (CV) of the scoliotic curves. The influence of the distance from the medial pedicle cortex to the spinal cord on t-EMG threshold intensity was also investigated at the apex segment. Whether the t-EMG stimulation threshold depends on pedicle bony integrity or on the distance to neural tissue remains elusive. Studying pedicle screws at the CC and CV at the apex segments of scoliotic curves is a good model to address this issue because the spinal cord is displaced to the CC in these patients. A total of 23 patients who underwent posterior fusions using 358 pedicle thoracic screws were reviewed. All patients presented main thoracic scoliosis, with a mean Cobb angle of 58.3 degrees (range, 46-87 degrees). Accuracy of the screw placement was tested at surgery by the t-EMG technique. During surgery, 8 screws placed at the CC showed t-EMG threshold values below 7 mA and were carefully removed. Another 25 screws disclosed stimulation thresholds within the range of 7 to 12 mA. After checking the screw positions by intraoperative fluoroscopy, 15 screws were removed because of clear signs of malpositioning. Every patient underwent a preoperative magnetic resonance imaging examination, in which the distances from the spinal cord to the pedicles of the concave and convex sides at 3 apex vertebrae were measured. Postoperative computed tomographic scans were used in all patients to detect screw malpositioning of the final 335 screws. According to postoperative computed tomographic scans, 44 screws (13.1%) showed different malpositions: 40 screws (11.9%) perforated the medial pedicle wall, but only 11 screws (3.2%) were completely inside the spinal canal. If we considered the 23 screws removed during surgery, the true rate of misplaced screws increased to 18.7%. In those screws that preserved the pedicle cortex (well-positioned screws), EMG thresholds from the CC showed statistically significantly lower values than those registered at the CV of the deformity (21.1 ± 8.2 vs 23.9 ± 7.7 mA, P < 0.01). In the concave side, t-EMG threshold values under 8 mA should be unacceptable because they correspond to screw malpositioning. Threshold values above 14 mA indicate an accurate intrapedicular position with certainty. At the convex side, threshold values below 11 mA always indicate screw malpositioning, and values above 19 mA imply accurate screw placement. At the 3 apex vertebrae, the average pedicle-spinal cord distance was 2.2 ± 0.7 mm at the concave side and 9.8 ± 4.3 mm at the convex side (P < 0.001). In well-positioned screws, a correlation between pedicle-dural sac distance and t-EMG threshold values was found at the concave side only (Pearson r = 0.467, P < 0.05). None of the patients with misplaced screws showed postoperative neurological impairment. Independent of the screw position, average t-EMG thresholds were always higher at the CV in the apex and above the apex regions, presuming that the distance from the pedicle to the spinal cord plays an important role in electrical transmission. The t-EMG technique has low sensitivity to predict screw malpositioning and cannot discriminate between medial cortex breakages and complete invasion of the spinal canal.