How to improve the safety of bicortical pedicle screw insertion in the thoracolumbar vertebrae: analysis base on three-dimensional CT reconstruction of patients in the prone position

Forty adults were selected to participate in the study: twenty men and twenty women between ages of 21 and 76 with a mean age of 53.4 years old. All participants had no thoracolumbar deformities, major vascular malformations, and anterior thoracolumbar vascular lesions, and none of them had any history of retroperitoneal surgery or thoracolumbar surgery.

Computed tomography was carried out by using PHILIPS brilliance iCT 256-row spiral CT, with PHILIPS image post-processing system. All subjects were injected with iodine contrast agent, and statistics were analyzed by using SPSS 22.0 statistical software.

The thoracolumbar region was reconstructed in three dimensions CT, with the patients in the prone position and the supine position respectively. In order to obtain a clearer image, iodine contrast agent was injected during the scanning process for angiography. The collected images were reconstructed by Philips ICT image post-processing workstation.

Fig.1 showed the axial image of the optimal pedicle screw trajectory plane. The following measurements were acquired: the angle of great vessels relative to the vertebral body (∠AOY/∠VOY; Fig. 1), and the shortest distance of between the great vessels and vertebral body in this direction. (AVD/VVD),

The experiment was designed to obtain an axial image of the optimal pedicle screw trajectory plane, which was perpendicular to the posterior plane of the vertebral body. In order to facilitate observation and measurement, the reference line was marked as axis X, axis Y, axis Z, and origin O. The X and Z axes passed through the posterior plane of the vertebral body, X axis passed through the center of the pedicle, and the Y axis passed through the midline of the vertebral body. (Fig.1, Fig.2).

On the axial plane, the Y axis was marked as 0°. The ∠AOY/∠VOY of the great vessel refers to the average angle between the tangent point on both sides of the vessel and the origin O relative to the Y axis. The AVD/VVD was measured in the direction of the angle of ∠AOY/∠VOY. (Fig.3).

In the prone and the supine positions, the distance from the aorta to the vertebral body of the T9 ~ L3 segment decreased at first and then increased, as shown in Fig.4. In the prone position, the minimum AVD was at T12 which was 3.39 ± 0.99 mm, followed by L1 which was 3.70 ± 1.43 mm; the maximum AVD was at T9 which was 5.94 ± 1.73 mm. In the supine position, the minimum AVD was at T12 which was 3.18 ± 0.68 mm, followed by L1 which was 3.70 ± 0.83 mm; the maximum AVD was at L3 which was 5.74 ± 2.65 mm. The AVD increased more in the prone position than in the supine position, and the AVD values measured at the T9, T10, T11, L2 vertebral bodies differed significantly between the prone and the supine positions (P < 0.05; Table 1).

In the prone position, the aorta of the T9-L3 segment tended to shift from the anterolateral side of the vertebral body to the anteromedial side of the vertebral body, as shown in Fig. 5, getting closer to the Y axis. At first, the ∠AOY gradually decreased from the level of T9 vertebrae (19.47 ± 10.49°) to the level of T11 vertebrae (11.85 ± 8.60°). It then increased, and a short peak appeared at the level of T12 and L1 (14.16 ± 6.84 °, 14.11 ± 8.71 °), and then decreased to the level of L3 (5.30 ± 5.68 °). In the supine position, the relative angle showed a similar result, the aorta gradually approaching the midline of the vertebral body. (Fig.5). Except for T12 and L1, there were significant differences between the angles of AOY in the prone and the supine positions (P < 0.05) (Table 1).

In the prone and the supine positions, the distance between IVA and vertebral body of L1 ~ L3 segment decreased gradually (Fig.6). However, the VVD values measured at the L1–L3 vertebral bodies didn’t differ significantly between the prone and supine positions (P ≥ 0.05; Table 2).

In the prone and the supine positions, the IVA of the L1-L3 segment was located in front of the right anterior side of the vertebral body (Fig. 7). However, it was limited to the 20°–30°position near the midline of the vertebral body (Y axis). There was no significant difference in the values of ∠VOY between prone position and supine position (P ≥ 0.05; Table 2).

The intra-observer reliability analysis on measurements of the distance and angle showed that the correlation coefficients of AVD/VVD and∠AOY/∠VOY are 0.95 and 0.93 respectively, representing excellent reliability.

It has been the focus of scholars to enhance the pullout force and internal fixation stability of pedicle screw. Clinicians wisely use bone cement augmentation, cortical bone channels, expandable pedicle and novel screw fixation. These new techniques have achieved satisfactory clinical results for the osteoporosis patient [8, 15‐18]. But for patients with non-severe osteoporosis, especially those in the middle-aged and elderly with a certain amount of bone loss, surgical techniques that increase the strength of internal fixation may be a better option, particularly in increasing the diameter of the pedicle screw, the depth necessary for insertion, and the insertion angle show an advantage [19, 20].

Studies have shown that when the screw is inserted in the anterior cortex of the vertebral body but not penetrated, the fixation strength can be increased by 16%, and the anterior cortex can be broken through, which can increase the pedicle screw pullout force by 60% and fixation strength by 20–25% [21‐24]. Bicortical anchorage increases the length of screw insertion and the stress is dispersed between the two cortical bones so that the fixation strength is significantly higher than that of cancellous bone [4, 5]. Its reliable fixation strength has also served in adolescent idiopathic spinal deformity or spinal deformity.

Huitema et al. [25] proved a substantial difference in the position of the aorta relative to the spine in the prone and in the supine position (range, T4-L2), while Vaccaro et al. [26] demonstrated substantial mobility of the great vessels in different positions (range, L4-S1). We also verified the results of the aorta, and we found that the AVD of T12 is the shortest distance both in the prone and the supine positions. Considering anatomical factors, the thoracic aorta extends into the abdominal aorta from the aortic sac of the diaphragm, which is mostly located at the T12-L1 positions and is close to the vertebral body. Thus, the aorta and vertebral bodies are fixed at T12 and L1 level and will not change due to changes in body positions. Compared with the aorta, we believe that the mobility of IVC has no obvious changes in different body positions.

Sarwahi et al. [27] claimed that anterior/anterolateral protrusion is less than or equal to 4 mm on CT poses no significant risk of impingement and can be considered safe. In gross anatomy, 23 misplaced screws do not endanger any structures and the distance they protruded are less than 4 mm on CT scan [27]. Because of the large distance between the large vessels and the vertebral body in the prone position, we evaluated and measured the safe distance in the supine position with CT. The shortest distance between the aorta and the vertebra is 3.18 ± 0.68 mm at T12, and the shortest distance between the IVA and the vertebra is 5.50 ± 2.17 mm at the L3 level. Considering the large distance in the prone position and the poor visualization of soft tissue on CT images, we conservatively believe that it is safe for the protruding tip of the screw to be less than that of 3 mm. Due to individual differences, we recommend that the actual safe distance between the great vessels and the vertebral body can be measured according to our method before surgery. In addition, imaging examination shows that labial hyperosteogeny may occur on the upper and lower margin of the vertebral body in elderly patients, pushing the anterior vertebral vessels to the front of the vertebral body (Fig. 2a), increasing the safe distance between the blood vessel and vertebral body.

When evaluating misplaced screws in contact with blood vessels, the protruding tips of some of the screws are too long, which puts more degrees of impingement on the blood vessels and is more likely to cause chronic vascular injury [28]. There will also be contact between the screw and the blood vessel after the bicortical fixation. However, patients with pedicle screws in contact with major vessels may not necessarily suffer adverse sequelae [29‐31]. We believe that if the tip of the screw can be controlled within the safe range, it will cause a lesser degree of screw impingement into the blood vessels and is unlikely to cause vascular damage [28].

Due to individual differences, some segmental vertebrae are close to blood vessels, and there is no obvious safe distance. However, simulating screw placement on preoperative supine and prone CT images can find the appropriate transverse screw angle (TSA), and the screw direction can avoid great vessels completely. As shown in Fig. 8, postoperative CT image of the patients with bicortical fixation presented that the great vessels of the L2 vertebral body can avoid the direction of the screw axis. Liu et.al [32] summarized the appropriate TSA of each pedicle of L1-L4. However, there is still an error in the insertion point of pedicle screw between the preoperative evaluation and the actual operation, which leads to a greater error in TSA. This requires surgeons to build their own preoperative models according to habits and to plan the correct TSA range to reduce errors.

This study also had certain limitations, such as a small sample size and inevitable measurement error, although we have chosen angiography CT to minimize errors. We observed that at the level of T8-T12, the boundary between the IVC and liver tissue passing through the hepatic vena cava sulcus is not obvious. However, there is a sufficient safe distance between it and the vertebral body, which will not affect the bicortical anchorage at all.(Fig.2) In addition, we also agree that bicortical anchorage is not suitable for implantation of L4-L5 vertebrae [32], and the above relevant data were not collected in the experiment.