Three-dimensional morphological analysis of the thoracic pedicle and related radiographic factors in adolescent idiopathic scoliosis

Pedicle screw (PS) fixation at multiple levels is commonly used worldwide in spinal surgery for three-dimensional (3D) correction of adolescent idiopathic scoliosis (AIS) [1, 2]. However, a misplaced PS can result in neurological, vascular, and visceral complications [3]. The pedicle morphology in AIS is narrower than that in a normal spine and varies highly among the different curve types [4]. Therefore, high accuracy in PS placement is required [5‐7], and preoperative planning of PS placement is essential [8]. In particular, PS insertion in the apical vertebra (AV) region on the concave side provides a positive correction effect; therefore, it is preferred whenever possible [9]. Right-left side-to-side differences in the pedicle dimensions at and near the apex in AIS patients are greater than those in healthy individuals [10], and PS insertion cannot be performed because the pedicle width is small, particularly on the concave side [11].

Recently, the use of computed tomography (CT) images has become common and important for the preoperative assessment of PS insertion [12]. However, in previous reports, the pedicle diameter was often evaluated using a single-slice CT image reconstructed based on the “pedicle axis”, which was subjectively set by the radiologist [13‐15]. In AIS with a complex 3D deformation, the pedicle axis differs significantly in each vertebral body. Therefore, the error in pedicle diameter measurement is large and depends on how the radiologist has fixed the slice position in the pedicle. Previous studies have investigated the use of cadavers in obtaining accurate pedicle measurements from 3D images [16‐18]. However, there is no report of accurate evaluation of the pedicle diameter by reconstructing the image of each vertebral body and the plane orthogonal to the left and right pedicle axes using a CT-based 3D image.

This study aimed to investigate the laterality of the thoracic pedicle morphology in the AV region of the main thoracic curve (MTC) in AIS patients using a large series of 3D-CT reconstructed images. The study also aimed to identify the radiographical factors associated with the laterality ratio of the pedicle morphology in the AV region of the MTC. We defined and used the term “laterality ratio” as a relative, rather than absolute, value. The laterality ratio was used as a measurement of the laterality of the pedicle diameter within an individual patient with high accuracy. We defined the “laterality ratio” as “the pedicle width index: the ratio of the left and right pedicle diameters of the same vertebral body,” as previously described [19].

A total of 57 patients were recruited for this study. There were 49 women (86.0%) and 8 men (14.0%), with a mean age of 16.3 ± 2.4 years. The mean height was 160.0 ± 6.7 cm and the mean weight was 49.2 ± 7.9 kg. The mean body mass index was 19.3 ± 2.4 kg/m². Lenke types 1 and 2 were observed in 25 (43.9%) and 32 (56.1%) patients respectively. Lumbar modifiers A and B-C were present in 45 (78.9%) and 12 (21.1%) patients, respectively. Risser grades 0–3 and 4–5 were observed in 7 (12.3%) and 50 (87.7%) patients, respectively. The following variables were measured for the MTC: the mean Cobb angle was 58.5 ± 8.9°; the level of the upper NV varied between T4 and T8; the level of the AV varied between T6 and T12; the length of the curves was 7.7 ± 1.0 vertebrae; the mean AVT was 57.1 ± 17.2 mm; the mean AVR was 19.4 ± 6.7°; and the mean TK was 15.5 ± 6.7° (Table 1).

For the 57 observed AIS patients with primary right thoracic (Lenke types 1–2) curves, a total of 342 thoracic vertebrae were identified in the 3D-CT images constructed with Synapse Vincent®, forming a database of 684 corresponding (left and right) pedicles.

Accurate measurement of the pedicle diameter in AIS surgery is important to prevent complications from PS malposition [3, 23] and to obtain a PS fixation with the target pedicle occupancy rate of > 80% [24, 25]. In particular, PS insertion in APEX±1 has been reported to have a positive effect on curve correction [9], and the measurement of the pedicle diameter in APEX±1 based on the AV level is highly critical for surgical planning. While the AV level differs by the curve types, there has been no pedicle morphological analysis based on the AV position in each patient. Furthermore, when evaluating pedicle diameters on CT images, if the pedicle morphometry is not performed on the plane that accurately determines the pedicle axis, the risk of overestimating the true pedicle diameter increases [26]. Therefore, in this study, we measured pedicle diameter on 3D-CT images, and we measured the isthmus of the pedicle orthogonal to the pedicle axis.

Pedicle diameter differences were significantly larger by a maximum of approximately 1.6 mm on the right side than on the left side in APEX±1, and the laterality ratio of the pedicle diameter tended to be large at Lt.w/Rt.w and small at Lt.H/Rt.H. However, the Lt.w/Rt.w of cancellous bone in NV ± 1 was 1.38, and we should not have discussed cancellous bone widths in APEX±1 using NV ± 1 as the control region. The reason may be that the pedicles of the upper thoracic vertebrae of AIS are often atypical and do not have cancellous bone [12], and the cancellous bone widths at NV ± 1 vary extensively. The predictors of Lt.W/Rt.W were the level of the AV and AVR. Interestingly, this new finding indicated that more proximal AV level or smaller AVR was associated with smaller Lt.W/Rt.W (i.e., larger laterality ratio of the pedicle diameter of the cortical bone width in APEX±1). To examine the laterality of the pedicle diameter in each patient, we referred to a previous study [19] for the definition of the “laterality ratio”. Furthermore, although there were no correlations between height, weight, and gender with pedicle width in this study, the thoracic pedicle diameter is generally influenced by height, weight, or gender in individuals without AIS [27, 28]. These reports also showed the reason a left-right difference comparison was not conducted using the absolute value of the pedicle diameter. Moreover, there is no consensus on whether the left-right ratio or the left-right absolute value difference should be used to compare the left-right morphology of the pedicle diameter. The closer the level of AV is to T4 or 5, the smaller the pedicle width is [4]. Nevertheless, we found that the closer the level of AV was to T4 or 5, the greater the value of Rt.W-Lt.W was (Supplementary Table 1A, 1B, and 1C). Therefore, we considered that the use of Lt.W/Rt.W to discuss the laterality of the pedicle diameter in this study was appropriate.

The most important determinant of the PS diameter is the width of the cortical bone [27]. Since a pedicle allows 15–25% enlargement [23, 28, 29], for a PS insertion of 4.0 mm, the outer diameter should be at least > 3.2 (4.0/1.25) mm. Our data suggested that 15 (26.3%) patients, presenting an APEX±1 pedicle on the concave side, would sustain a fracture if a screw for the thoracic spine was inserted. For patients in whom the PS could not be inserted at the concave side of APEX±1, we might consider using hook or sublayer wiring [30, 31]. Convex side correction with PS inserted into the convex side might also be considered a useful correction method [27]. Moreover, the in-out-in insertion techniques could be considered if insertion of the screw into the pedicle is determined to be challenging owing to the small width of the pedicle. However, while the in-out-in insertion technique minimizes the risk of nerve damage, sympathetic trunk damage might occur [16]. This technique was also associated with reduced pull-out strength compared to PS placed using an anatomical orbit [32]; thus, inserting a pedicle screw within the permissible range of pedicle enlargement is preferred.

In previous evaluations of the mean value for each vertebral body level, the difference between the left and right pedicle diameters of the middle and lower thoracic vertebrae was within 1.0 mm in cortical pedicle height, cancellous pedicle height, cortical pedicle width, and cancellous pedicle width [11, 23]. However, a new finding from this study showed that the left and right pedicle diameters significantly differed by > 1.0 mm. In particular, 41 (71.9%) patients had pedicle width differences > 1.0 mm, which are important for planning PS insertion. The reason might be that the recruited participants had only AIS types 1 and 2. Additionally, the left and right pedicle diameters in APEX±1 was compared based on the AV, and the measurement was performed on 3D-CT reconstructed images.

Two notably major findings from this study could be useful for spine surgeons. First, predicting pedicle morphology from the AV level is possible without preoperative CT. Second, at APEX±1, laterality of the cortical pedicle width on the opposite side of the AVR direction might exist (e.g., the pedicle width on the opposite side of the AVR direction is relatively large).

This study did not suggest the need for preoperative CT for the prevention of PS misplacement in AIS patients. Currently, both intra-operative navigation and, to an even greater extent, CT-based preoperative planning, are extremely useful tools. However, CT-based preoperative planning did not significantly reduce PS misplacement [33], and their use could be limited to selected cases. Therefore, understanding differences in pedicle morphology is considered even more important for the surgeons to prevent PS misplacement in patients who do not undergo preoperative CT.

Based on the results of this study, we concluded that greater vertebral body rotation in APEX±1 was associated with smaller laterality ratio (i.e., the larger Lt.W/Rt.W). In AIS patients, it was previously reported that larger curve magnitude of the coronal plane was associated with a larger increase in pedicle dysplasia at the apex and concave side of the curve [12, 34]. However, in this study, the Cobb angle of MTC was not a related factor of Lt.W/Rt.W in APEX±1. In other words, our results suggested that the vertebral rotation, an axial evaluation, affected Lt.W/Rt.W more significantly than the curve magnitude, a coronal plane evaluation. This phenomenon could be related to a pathoetiological mechanism, by which the AVR and spinous processes are deformed to compensate for each other in the axial plane [35]. Another possible explanation was that the pedicle diameters were evaluated not as the absolute value but as the left-right ratio of a particular vertebral body in the multiple linear regression analysis of this study. Therefore, our results might differ from those of previous reports [12, 34].

This study had several limitations. First, computerized measurements might not capture the entire natural biological variability of the pedicle shape as well as every pathological shape. However, the advantage of computerized measurements is that they allow the extraction of additional pedicle morphometric parameters and thus, can reliably support spine surgeons in selecting proper screw sizes as well as inserting trajectories during preoperative planning for PS placement procedures. Second, PS diameters through the isthmus might differ by several millimeters, depending on the selection of the insertion points and differences in the pedicle axis and PS insertion axis, because the surgeon did not always insert the PS in the same trajectory as the pedicle axis when considering rotation correction and the PS placement sequence.

In AIS with Lenke type 1 or 2, 26.3% of the pedicles in APEX±1 on the concave side were too small for a 4-mm-diameter screw even with an expansion. The right and left pedicle diameter ratio in APEX±1 for the height or width of the cortical or cancellous bone ranged from 0.7 to 0.9. There was a notable difference in the pedicle diameter ratio between the left and right pedicle widths of the cancellous bone (Lt.W/Rt.W). Furthermore, the pedicle diameter ratio between the left and right widths of the cortical bone was larger in patients with more proximal AV and smaller AVR. Although preoperative planning of each AIS patient remains important for surgeons, this study may help plan the selection of available instrumentation from information, other than the preoperative CT, that could predict the laterality of the pedicle morphology at APEX±1 from the preoperative curve type.