Introduction

The goals of thoracolumbar vertebral burst fracture treatment include restoring vertebral column stability and preventing deformity, neural canal decompression, and early mobilization [1, 7, 9, 18, 24, 28, 34]. Although anterior, posterior, and combined surgical approaches are all used for this purpose, the selection of the treatment approach is still controversial [22]. The posterior approach is used more frequently among spinal surgeons because of its easy application, reduction of bleeding, and small incision area. In contrast, the application of short- or long-segment pedicle screws is controversial. Despite the advantages of long-segment instrumentation, such as tighter fixation and better canal recovery, it also results in a motionless spine due to fixing more segments [1, 18, 24, 28]. In recent years, short-segment instrumentation with the pedicle screws introduced one level down and one level up from the fractured vertebra has become the preferable surgical method because of its ease of application, use of less surgical fixation material, reduction of blood loss, and smaller incision field [1, 2, 13, 18, 20, 22, 24, 28, 31, 33]. However, disadvantages of this method, such as inadequate long-term reduction, instrumentation insufficiency, and increases in kyphosis and pain, have also been reported [1, 2, 5, 7, 14, 17–20, 32, 34]. Osteoporosis, inadequate support of the anterior column, insufficiency of fixation points, and the application of laminectomy have been affirmed as potential reasons for insufficiency. To prevent such insufficiency, alternative procedures [1, 6, 7, 26, 27, 29] supporting the anterior column have been suggested, such as vertebroplasty, kyphoplasty, or screw application for burst vertebrae. This method results in stronger biomechanical stability of the anterior column by forming a more segmental structure [4, 13, 15, 18].

The purpose of this study was to evaluate the efficacy of the placement of an additional screw at the fracture level regarding the correction of deformity, maintenance of the correction, and prevention of fixation failure in thoracolumbar burst fractures.

Methods

The study included 70 patients who underwent short-segment stabilization because of the diagnosis of thoracolumbar (T11-L2) burst fracture between 2008 and 2012. Only patients who were neurologically intact (ASIA-E) with burst fractures according to the Denis [9] classification were included. The patients were evaluated according to their age/gender, trauma etiology, and fracture level (Table 1). The follow-up periods for patients in both groups ranged from 9–31 (average: 26.5) months. The stability of at least one pedicle was confirmed; patients in whom both pedicles were fractured were excluded from the study. All patients with poor bone mineral density (T score ≥2.5) or instrumentation failure (screw breakage in two patients in group 1) were also excluded. Three different surgeons operated on the patients. In 35 patients (group 1), a pedicle screw was placed only one level down and one level up from the fracture level; in another 35 patients (group 2), a screw was placed at the fracture level in addition to the short segment. Papers that referred to screws at the fracture level improving construct stiffness influenced our surgical strategy. After 2010, we placed screws at the fracture level in all appropriate cases. In three patients, only one screw was inserted at the fracture level in group 2.

Table 1 Demographic properties of the patients according to their age-gender, etiology, fracture level, and Denis burst fracture type
Full size table

Radiographic review

The preoperative and most recent postoperative follow-up radiographs and CTs of the patients were evaluated in terms of the sagittal index, kyphosis angle (Cobb), ratio of canal compromise, and anterior vertebral height. The sagittal index was calculated by measuring the angle that crosses lines drawn from the upper and lower endplates of the traumatic vertebra, as defined by Farcy et al. [12]. The local kyphosis (Cobb) angle was calculated by measuring the angle between the upper endplate of the vertebra one level up and the lower endplate of the lower vertebra [15]. The amount of retropulsion (spinal canal compromise) was measured using computed tomographic (CT) scans, and the percentage of spinal canal compromise was expressed as follows [23]:

Percentage of spinal canal compromise:

$$ \mathrm{a}=\left(1-\mathrm{x}/\mathrm{y}\right)\times 100 $$
a:

Percentage of canal compromise

x:

Mid-sagittal diameter of the spinal canal at the level of injury

y:

Average mid-sagittal diameter of the spinal canal (one level above and one level below the level of injury)

The anterior body height of the traumatic vertebra was evaluated by taking the percentage of anterior body height compression as a reference by measuring the anterior heights of the one-level-up and one-level-down vertebrae using the method of Mumford et al. [25]. The preoperative and follow-up radiographs were evaluated.

Surgical technique

Surgery was indicated in the patients in whom the sagittal index was above 15°, the Cobb angle was above 10°, the ratio of canal compromise was greater than 25 %, and the anterior vertebral height was less than 50 % of the posterior vertebral height, according to the method of Mumford et al.

A majority of the patients underwent surgical intervention within 72 h. The patients were evaluated based on preoperative radiograms, thin-slice CT and MRI, and radiographs and CT in postoperative controls. Their pedicle levels were scanned using thin-slice CT. Before surgery and in the controls, radiographic angles were measured.

Two different (TIPMED-Izmir, Turkey, HIPOKRAT-Izmir, Turkey) instrumentation methods were employed. Fracture reduction and indirect decompression of the spinal cord were achieved by applying distraction and producing an appropriate contour in the rod. Transverse connections were used in each case. No laminectomies were performed. Kyphosis was corrected only through postural reduction and cantilever reduction of the rods in the screws. Autografts and allografts were used for posterolateral bone fusion. The patients were mobilized with a corset on the 2nd day after surgery, and it was recommended that they wear the corset for 1 month.

Statistical analysis

The SPSS (Statistical Package for the Social Sciences) 18.0 software package was used for the statistical analysis of the data. Categorical measurements were summarized as numbers and percentages, and numerical measurements were summarized as the average and standard deviation (minimum-maximum). Chi-squared statistics were employed in the comparison of categorical measurements between groups, and independent t-tests were employed for the comparison of numerical measurements between groups. The statistical significance level was taken as 0.05 in all tests.

Results

Age-gender

The age range of the patients in group 1 was 18–70 (average age: 39.2) years, and the female:male ratio was 10:25. The age range in group 2 was 18–61 (average age: 40.4) years, and the female:male ratio was 8:27 (Table 1).

Etiology

All patients experienced high-energy trauma. Falling from height was the most common reason for injury, presented by 63 % of the patients in both groups. The etiology of the injury was a traffic accident in 12 patients in group 1 and 9 patients in group 2, and the impact of a hard body was observed in 1 patient in group 1 and 4 patients in group 2 (Table 1).

Fracture level

Thoracolumbar fractures between T11 and L2 were included in this study. L1 fracture was sustained by 51 % of the patients in the first group and 54 % in the second group. When groups 1 and 2 were compared according to the fracture level, the results were as follows: T11:3/2, T12:7/8, and L2:7/6 (Table 1).

Denis classification

B-type fractures were observed in 20 patients (57 %) in each group. When groups 1 and2 were compared according to Denis type, the results were as follows: Denis A: 5/8, Denis C: 6/5, and Denis D: 4/2.

No statistically significant differences were found between the groups in terms of age, gender, etiology, fracture localization, or Denis classification (Table 1).

Sagittal index

The average sagittal index was 20.5° preoperatively and 11.4° in the controls in group 1, while it was 19.8° preoperatively and 7.4° in the controls in group 2. The results showed a statistically significant difference (p < 0.001) between groups 1 and 2 in terms of the sagittal index in the controls (Fig. 1).

Fig. 1
figure 1

The comparisons of radiological measurements of the canal comprise the ratio, vertebral height, and sagittal index between the two groups (*p value: 0.002, **p value: 0.001)

Full size image

Local kyphosis (Cobb) angle

The average Cobb angle was 17.4° preoperatively and 10.5° in the controls in group 1, while it was 17.3° preoperatively and 5.4° in the controls in group 2. No statistically significant differences were observed between the groups in terms of the preoperative evaluation. The difference in the Cobb angle between groups 1 and 2 was statistically significant (p = 0.006) in the controls (Fig. 2).

Fig. 2
figure 2

Kyphosis angle in groups 1 and 2 (***p value: 0.006)

Full size image

Ratio of canal compromise

While the ratio of canal compromise was 30.7 % in group 1, it was 14.7 % in the controls. The ratio was 32.1 % in group 2 and 12 % in the controls. The ratio of canal compromise was of limited significance (p = 0.189) in group 2 in the controls (Fig. 1).

Anterior vertebral height

While the anterior vertebral height was 36.3 % preoperatively and 17.6 % in the controls in group 1, it was 37 % preoperatively and 10 % in the controls in group 2. These results were statistically significant (p = 0.002) in both groups (Fig. 1).

Discussion

The ideal treatment of thoracolumbar burst fractures remains a matter of discussion. Posterior transpedicle instrumentation is the most frequently applied surgical treatment for these fractures because of its low morbidity and comorbidity [3, 7, 8, 10, 19].

In the current study, two patient groups who underwent short-segment posterior fixation were compared. The demographic, clinical, and radiologic properties of the two groups were matched as closely as possible. Homogeneity was confirmed, as no statistical significance was observed in terms of the sagittal index, kyphosis angle, ratio of canal compromise, or anterior vertebral height at preoperative evaluation. These parameters were found to be statistically significant in the long-term control period, indicating that short stabilization with the application of a pedicle screw at the fracture level results in a greater correction of kyphotic deformity, an increase in anterior vertebral height, and a decrease in the sagittal index. In the controls, the sagittal index was 11.4° in group 1, while it was 7.4° in group 2; the kyphosis angle was corrected by 6.9° in group 1 and by 11.9° in group 2. The anterior vertebral height was increased from 36.3 % to 17.6 % in group 1 and from 37 % to 12 in group 2.

In recent years, insufficiency of the implants and loss of correction have been reported as the most significant disadvantages of short-segment instrumentation by some authors [1, 17, 20, 27, 28, 32, 34]. Because residual kyphotic deformity generates high anterior vertebral stress on pedicle screws, failure, dislocation, and disconnection of screws due to overload on the instrument are insufficiencies that are frequently observed in short-segment fixation [7, 8, 19, 24, 28, 34]. Increasing the fixation level decreases the chance of this insufficiency by reducing the stress on each pedicle. However, it also decreases the protective advantage of mobile segments compared with short-segment instrumentation [1, 18, 20, 24, 28, 34]. Some authors have reported successful results when using short-segment instrumentation in long-term controls [21, 24, 30, 33]. In cadaver studies conducted by Mahar et al., adding pedicle screws at the fracture level in addition to short-segment pedicle fixation in burst fractures was shown to significantly increase spinal stability [18]. According to the hypothesis of Guven et al., intraoperative fracture reduction and correction of sagittal deformity can be easily achieved via the placement of a screw at the fracture level [15]. Anekstein et al. reported that screws placed into fractured vertebrae can maintain the burst vertebra and separated pedicle [4]. Gelb et al. found that thoracolumbar fractures can be successfully treated with short-segment pedicle instrumentation [13]. The reported advantages of this method include the protection of more mobile segments and reduction of donor field complications, the operation duration, and blood loss. Some researchers have stated that short-segment pedicle instrumentation is the best choice in terms of flat back syndrome and loss of lumbar lordosis [3].

In cases of neurological deficit, the combined anterior and posterior approach is a treatment option that achieves complete kyphosis correction, immediate stability, and complete spinal canal decompression [28]. However, this technique has not been widely accepted in neurologically intact patients because it requires a more invasive surgical procedure and increases operative time, blood loss, and morbidity [16].

Another alternative that is increasingly being used in recent years is to place a screw within the fractured vertebra [4, 13, 15, 18]. According to Mahar et al., the application of limited posterior segmental instrumentation in thoracolumbar burst fractures is a method that achieves short-segment fixation. Segmental construction using a pedicle screw in the fractured vertebral body has been shown to be more reliable and more corrective in terms of biomechanical stability compared with non-segmental construction [18]. Axial torsion stability has been shown to be approximately 2 times greater in biomechanical tests in cadaver models. Incremental increases in flexion and extension stability and in lateral bending have been achieved, although they have not been statistically significant.

Screws placed within the fractured vertebra can hold a mechanically burst vertebra and separated pedicles together. In short-segment fixation performed without placing a screw into the fractured vertebra, cavities occurring inside the fractured vertebra after restoration will eventually lead to a loss of correction. Guven et al. observed increments of compression, anterior vertebral height, or the kyphosis angle in long-term controls for long- and short-segment fixation [15]. However, these increments are more substantial when a screw is not placed within the fractured vertebra, and this difference is statistically significant. In this study, screws placed at the fracture level were found to generate a mass effect and to prevent vertebral collapse. According to the results of Guven et al., better correction is achieved in short-segment applications combined with a screw placed at the fracture level compared with short-segment fixations in which no screw is used. The recovery of the kyphosis angle and anterior vertebral height is superior in long-segment fixations in which a screw is applied to the fracture level.

In many studies, the correction of the kyphosis angle observed in the early postoperative period is decreased in long-term controls [5, 7, 13, 18, 20, 21, 34]. Carl et al. [5] reported the first postoperative kyphosis correction to be 7°, while Cho et al. [7] achieved a correction of 6°. In both studies, the early kyphosis correction observed in the postoperative period was lower than in the controls. Similarly, McNamara et al. [21] reported an initial loss of the kyphosis correction of 9° when using the non-segmental fixation technique. McLain et al. [20] found that progressive deformity developed within 6 months postoperatively in a majority of patients with residual anterior column instability. Gelb et al. [13] used the fractured vertebra as an intermediate fixation point in 74 % of patients. The loss correction observed in Gelb's study was the same as in other studies. In the current study, we only evaluated the most recent postoperative radiological measurements.

According to Gelb et al., the success of short-segment pedicle instrumentation depends upon multiple factors [13]. Current instrumentation with improved metallurgy and screw-to-rod locking mechanisms has undoubtedly contributed to the decrease in instrumentation failure. Advances in surgical techniques have also been instrumental in the success of the construct. Early surgical intervention promotes easier postural reduction. We advise precontouring of the rod, providing a three-point bending force as the apex of the rod engages the intermediate fractured vertebra, similar to the concept of the spinal rod-sleeve method previously described by Edwards and Levine [11]. The intermediate vertebra could be engaged directly as the rod contacts the posterior lamina or through use of a screw into one or both of the pedicles of the fractured vertebra.

A limitation of this study was clinical data such as the VAS score and quality of life, which were absent in this study.

Conclusion

The purpose of surgical treatment of thoracolumbar junction burst fractures is to preserve the height and alignment of the vertebral body, decompress the spinal cord, enable early ambulation and rehabilitation with tight fixation, and prevent progressive deformity and neurologic deficits. While attempting to achieve these aims, the number of immobile segments should be limited by instrumenting as few vertebrae as possible.

Short-segment instrumentation using additional screws at the fracture level in thoracolumbar burst fractures is a proper surgical approach for obtaining clinically and radiologically successful results in terms of the sagittal index, kyphosis angle, ratio of canal occupation, and correction of collapse in the anterior body.