Common practice for surgical treatment of severe and rigid spinal deformities includes anterior release and fusion, one- or two-stage posterior fixation combined with spinal osteotomy, vertebral column resection (VCR), etc.. These treatments to correct and fix spinal deformities have been well accepted by spine surgeons. However, for severe, rigid, and angular spinal deformities, it is not easy to obtain desirable correction results using these approaches.
In recent years, some authors have reported the clinical use of posterior VCR (PVCR) for correcting spinal deformities in patients with spinal scoliosis or kyphosis [7, 8]. In particular, for patients with severe (Cobb angle >100°), angular (the angle changed significantly through a few spinal segments), and rigid (flexibility <10%) spinal deformities that are accompanied with significant coronal or sagittal decompensation, PVCR will be the best alternative . Severe spinal deformities are frequently accompanied by congenital or secondary pedicle deformities that are often located in the thoracic spine. When PVCR is performed to correct the deformities, the most crucial step is to insert screws in the pedicles, particularly pedicles in the apical region of the main curve in the thoracic spine.
Many factors have an impact on the accuracy and safety of thoracic pedicle screw placement. The rate of thoracic pedicle screw malpositioning during correction of spinal deformities is as high as 3.0–44.2%, and the rate complications related to pedicle screws is 0.9% [10–12]. For surgeons, especially in the process of applying PVCR, the preoperative reliable observation and analysis of each pedicle will be very helpful for the accurate insertion of screws and the final decision of fusion segments. Therefore, a quantitative description method or criteria for pedicles will help the preoperative plan, shorten the duration of operation and decrease the risks being caused by screw insertion. However, up to date, there are only few reports on the correlation between the inner cortical width of thoracic pedicles and the accuracy of thoracic pedicle screw placement.
The present study included a consecutive series of 56 patients with severe and rigid spinal deformities who underwent PVCR at a single institution between October 2004 and July 2010. A total of 1098 thoracic pedicles instrumented at T2-T12 were reviewed. We have proposed criteria for establishing a thoracic pedicle classification based on the inner cortical width of thoracic pedicles as determined on CT scans, and believe these criteria can be of clinical significance.
Selecting a measurement method to quantify relevant values for thoracic pedicles
When the accuracy of pedicle screw placement is evaluated, routine radiography is of little help to the surgeon to determine directly whether the pedicle has been penetrated. The accuracy of routine radiography for evaluating pedicle screw placement varies significantly, ranging from 10% to 83% . Routine two-dimensional CT scans can be easily affected by metal artifacts and cannot objectively quantify pedicle penetration. Three-dimensional reconstruction of pedicles on multi-slice spiral CT images is of great importance. By rotating and cutting VR images at random, this method can derive values indicating the inner cortical width of pedicles that are very close to the actual transverse diameter of the pedicle isthmus. It is therefore considered the gold standard [14, 15]. It is certainly more reliable than others. Therefore, in the present study, the thoracic vertebrae of scoliotic patients were scanned, and thin-slice VR and MPR of the thoracic pedicles were performed.
Some studies indicated that the gaps between thoracic pedicles of the upper and middle thoracic vertebrae and dural sac were small, whereas the gap between the pedicles of the lower thoracic vertebrae and dural sac are as large as 2 mm . Therefore, in the present study, when penetration of the internal and external osseous walls of thoracic pedicles was evaluated, it was measured in millimetres. On the basis of that evaluation, we proposed a method to classify thoracic pedicle penetration from the viewpoint of clinical significance. That is, the penetration of internal and external cortical walls of thoracic pedicles could be classified into three degrees: 0, no penetration; 1, thoracic pedicle penetration ≤2 mm; 2, thoracic pedicle penetration >2 mm. Degrees 0 and 1 penetration of thoracic pedicle walls indicate that the screws had been inserted into thoracic pedicles.
Whether the inner or outer cortical width of thoracic pedicles should be measured when studying the morphology of thoracic pedicles has been debated [3, 17–19]. In the present study, when measuring the cortical width of thoracic pedicles in the transverse plane, we measured the inner (rather than the outer) cortical width of thoracic pedicles because we believed that this method was more reliable for determining the pedicle screw diameter and for demonstrating the effects of thoracic pedicle instrumentation .
Characteristics and clinical significance of quantification classification of thoracic pedicles determined by their inner cortical width on CT images
Along with the pedicle classification method of Lenke et al.
, previous studies on pedicles focused on morphology. They lacked a quantification standard and were poorly consistent, making classification methods based on their findings rather subjective. Furthermore, such classifications gave little help to surgeons for predicting the accuracy of thoracic pedicle instrumentation with the free-hand technique before surgery. For all of these reasons, these classifications never gained popularity.
The study by Liljenqvist et al.
 noted that the angulations of spinal deformities in the coronal plane and rotation of the vertebral body were not significantly correlated with the accuracy of thoracic pedicle screw placement, a finding that has been gradually accepted. In the present study, after having meticulously studied 1098 thoracic pedicles of scoliotic patients on CT images and analysed the data, we noted that there was a statistically significant difference in the accuracy of thoracic pedicle screws between pedicles with and those without a channel. In the groups of thoracic pedicles having a channel, there was statistically significant difference in the accuracy of thoracic pedicle screws between pedicles with an inner cortical width of 1.1–2.0 mm and those with a width of ≥2.1 mm. According to the morphological classification of thoracic pedicles proposed by Lenke et al.
, there was a statistically significant difference in the accuracy of thoracic pedicle screws between their type D without a pedicle channel and the three other types (A–C) (P < 0.008). The statistical data of the current study and our clinical experiences indicated that, among many other factors affecting the accuracy and safety of thoracic pedicle screws, the angulations of the spinal deformities in the coronal plane and the rotation of the vertebral body only indirectly affected the accuracy and safety of thoracic pedicle screw placement by increasing the difficulty of the surgeries. The only crucial factors that had a direct impact on the accuracy and safety of thoracic pedicle screw placement and safety were (1) whether the pedicle had a channel and (2) the inner cortical width of the thoracic pedicles.
Therefore, we propose the following thoracic pedicle classification based on the inner cortical width of the thoracic pedicle on CT scans: type I thoracic pedicle, which has no channel and an inner cortical width of 0–1 mm; type II thoracic pedicle, which has a channel. Type II thoracic pedicles are then subclassified: type IIa, which has an inner cortical width of 1.1–2.0 mm, and type IIb, which has an inner cortical width of ≥2.1 mm. As the thoracic pedicle classification criteria have been quantified and the morphology of the thoracic pedicles is not taken into account, the consistency and reliance of this new thoracic pedicle classification method is an improvement. Comparisons on the accuracy of thoracic pedicle screws between the two classification methods showed that thoracic pedicle classification determined by the inner cortical width of the pedicle on CT images can help surgeons predict whether a screw could be accurately inserted into the thoracic pedicle—which was of clinical significance.
Regarding the distribution of different types of thoracic pedicle, types I, IIa, and IIb thoracic pedicles accounted for 17.67%, 16.03%, and 66.30%, respectively, of the total thoracic pedicles studied. The accuracies of pedicle screw placement in types I, IIa, and IIb thoracic pedicles were 35.05%, 65.34%, and 88.32%, respectively. Evaluation of thoracic pedicles prior to or during PVCR procedures is of significant importance. Type IIb pedicles of the thoracic vertebrae are suitable for screw insertion, and screw placement can be attempted in type IIa pedicles. In contrast, the accuracy of thoracic pedicle screw placement in type I pedicles is extremely low, particularly when the thoracic pedicles are adjacent to apical vertebrae or are above or below vertebrae that are to be resected. In these cases, PVCR should be selected cautiously or completely abandoned.
PVCR is an effective approach for treating severe spinal deformities. Whether the screw can be inserted into a morphologically abnormal pedicle is crucial for accurate pedicle screw placement. In the present study, we noted that the inner cortical width of the thoracic pedicles was the sole decisive factor for predicting the accuracy of thoracic pedicle screw placement. On that basis, we propose quantification classification criteria for thoracic pedicles that could help surgeons predict whether a screw could be inserted into a thoracic pedicle during PVCR, thus guiding the instrumentation of thoracic pedicles with the free-hand technique.