Triggered Electromyographic Threshold for Accuracy of Thoracic Pedicle Screw Placement in a Porcine Model

SJ Lewis, MD,
LG Lenke, MD,
BL Raynor, J Long, DVM,
KH Bridwell, MD,
KD Riew, MD,
AM Padberg, MS
Washington University School of Medicine,
Department of Orthopaedic Surgery, St. Louis, MO, USA

INTRODUCTION:
The use of pedicle screws is increasing in the thoracic spine. Misplaced thoracic pedicle screws may have significant implications if the spinal cord is injured. Triggered electromyographic (EMG) stimulation has been a valuable aid in determining appropriate placement of lumbar pedicle screws (Spine 1995;20:1585). The purpose of this study was to determine if established criteria used to assess pedicle screw placement in the lumbar spine were applicable to pedicle screws inserted in the thoracic spine in an animal model.

METHODS:
Four 120–150 lbs domestic pigs had 45 pedicle screws placed bilaterally in the thoracic spine. Pedicle screws were first inserted entirely in the pedicle (Group A). The medial pedicle wall was then removed and the screw was placed with no osseous bridge between the screw and the neural tissue, however, with no contact to the neural tissue (Group B). Finally, the screws were placed medial to the pedicle with purposeful contact to the nerve root and the spinal cord (Group C). Pairs of ½” subdermal needle electrodes were placed in the rectus abdominus and intercostal muscles for each instrumented spinal level. An ascending method of stimulation using constant current (mA) was applied to each screw to obtain a compound muscle action potential (CMAP). Threshold intensity and muscle(s) responding were recorded for each trial.

RESULTS:
In 38 of the 45 screws, there was a relatively consistent decrease in the triggered EMG response from Group A (mean 4.62±0.25 mA) to Group C (mean 2.08±0.18 mA) screws (p<0.05). There was little difference in the response obtained from Group A to Group B (mean 4.17±0.28 mA) screws (p>0.05). In Group B screws, 15 of 38 (40%) screws had thresholds greater than their respective A screw. In the case of 7 screws, the nerve root was felt to be injured by the technique of breaking the medial pedicle wall. In these screws, the responses obtained from the Group B (mean 6.32±1.21 mA) and C screws (mean 6.36±0.68 mA) were significantly greater than the responses obtained in Group A (mean 3.27±0.65 mA, p<0.05), suggesting that a greater threshold was required to obtain a response from the injured root.

DISCUSSION:
There was a mean 53.4 ± 3.8 % decrease in the EMG threshold for the screws with neural contact compared to the screws properly placed in the pedicle, however, the individual threshold for each screw was variable. Five of the 38 (13%) properly placed screws had thresholds below 3.0mA, with all but one of these values being higher than the corresponding C screw. There were five C screws (13%) placed in direct contact with the dura that scored greater than 3.0mA, with all of these thresholds lower than their respective A screw. Similarly, 28 of 38 B screws (74%) had thresholds greater than 3.0mA. Even though in an individual pedicle there was a consistent decrease between the A and C screws, we could not determine a cut–off trigger EMG level that would consistently differentiate properly from improperly placed pedicle screws. Furthermore, this method could not differentiate screws clearly in the pedicle with screws with medial pedicle wall breakthrough.

CONCLUSIONS:
We do not feel this animal model of root recordings to be reproducible in identifying misplaced thoracic pedicle screws. A more direct method of spinal cord monitoring must be established to provide the surgeon with early warning of the potential of neural injury in the placement of thoracic pedicle screws.