This study found that the vacuum phenomenon of the intervertebral disc adjacent to the fractured vertebral body before removal of the posterior pedicle screw was associated with secondary severe kyphotic deformity in patients after implant removal for thoracolumbar burst fracture. No such association was seen with old age, sex, the initial severe kyphosis, or initial severe wedge deformity. These findings should be helpful in the clinical setting because there has been little consensus to date on what constitutes routine implant removal in the context of healed fracture. Even though implant removal is probably common practice among spine surgeons after bone union, because it decreases the stiffness of the fixed segment and reduces the concentration of stress in the adjacent segments [15], the recurrence of kyphotic deformity is a recognized shortcoming of implant removal following posterior fixation, affecting 29%-43% of patients [9, 10]. Given that the prognostic factors for secondary kyphotic deformity have not been established yet, our findings can serve as a reference to assist physicians in deciding whether implant removal should be conducted or not.
In this study, final kyphotic deformity and loss of correction were set as the objective variables. To our knowledge, only two previous studies have investigated the predictors of kyphosis following implant removal in posterior pedicle screw fixation without fusion for thoracolumbar burst fracture [16, 17]. These studies set loss of correction alone as the objective variable, although the clinical significance of loss of correction following implant removal is as yet unclear in the literature [16, 17]. In contrast, final kyphotic deformity has previously been proposed to affect clinical outcomes. Kraemer et al. evaluated functional outcome in 24 patients with a minimum of 2 years of follow-up after thoracolumbar burst fracture without neurologic deficit using the SF-36 survey and the Roland scale and concluded that clinical outcomes in patients with kyphosis greater than 25° were poor [13].
In relation to loss of correction, this study demonstrated that initial severe kyphosis was not associated with such loss. Additionally, the kyphotic angle after implant removal was presumed to be affected by the intervertebral disc. Previous studies have reported varying outcomes for initial severe kyphosis. The reason for the lack of significant association between initial severe kyphosis and loss of correction might be because initial severe kyphosis was mainly affected by the initial severe wedge deformity of the fractured vertebral body. In a retrospective comparative study, Chen et al. did find an association between the two. In their study, they defined loss of correction as an increase of more than 5° in the kyphotic angle at the final observation compared with the kyphotic angle after posterior fixation. Initial severe kyphosis was evaluated using the percentage of the anterior vertebral body height with respect to the posterior vertebral body height of the fractured vertebra and the percentage of the anterior vertebral body height with respect to mean anterior height of the upper and lower adjacent vertebrae [16]. Also, in a comparative retrospective study by Aono et al., initial severe kyphosis at the time of injury was associated with loss of correction. In their study, they used an increase of more than than 10° in the kyphotic angle at the final observation compared with kyphotic angle after posterior fixation, and they evaluated initial severe kyphosis in the same way as in our study [17]. We decided to use an increase of more than 15° in our study based on a review article by Mazel et al. in which thoracolumbar burst fracture with kyphosis of 15° or more was considered to require correction [18].
This study revealed an association of the vacuum phenomenon of the intervertebral disc adjacent to the fractured vertebral body before removal of the posterior pedicle screw with secondary severe kyphotic deformity following implant removal for thoracolumbar burst fracture. The vacuum phenomenon may be the result of disc degeneration. Previous studies have shown that initial severe kyphosis resulted from loss of not only vertebral height but also disc height and that initial vertebral height at the time of injury was corrected after primary surgery and maintained after implant removal [19, 20]. Similarly, the fractured vertebral angle was maintained after implant removal in this study. On the other hand, the angle of the intervertebral disc adjacent to the fractured vertebral body at the time of injury increased after posterior fixation but was not maintained after implant removal. This loss of disc angle may be due to degeneration of the disc, which might be accelerated by traumatic damage at the time of injury. Similarly, Kanezaki et al. reported that the association of severe disc damage at the time of injury, as evaluated with magnetic resonance imaging (MRI), with kyphosis after implant removal [21]. While the outcome from their study is valuable, CT imaging is more convenient in daily practice than MRI. At our institution, MRI evaluation is indicated for patients with neurological deficit, not for all patients with thoracolumbar burst fracture. We believe our CT findings in this study can prove valuable for the many physicians who use CT in daily practice.
Old age, which was defined as age above 60 years, was not associated with kyphotic deformity in this study, in contrast with the results from Chen et al. In their study, patients were divided into recurrence and non-recurrence groups based on loss of correction. Mean age was significantly higher in the recurrence group (40.1±9.5 years) than in the non-recurrence group (35.3±8.1 years) [16]. They discussed that loss of correction was associated with patient age due to age-related loss of bone density. The definition of old age in the present study was more reflective of loss of bone density than in their study and age was not associated with kyphotic deformity in this study. The difference between these studies suggests caution is needed in interpreting the effect of age on secondary kyphotic deformity following implant removal in posterior pedicle screw fixation.
This study provides new criteria for physicians to decide whether or not to perform implant removal. Because final kyphotic deformity is suggested to cause poor clinical outcomes, physicians can cautiously consider the advantages and disadvantages of removing the pedicle screw for patients with the vacuum phenomenon apparent on CT examination [13]. Several studies have reported on stem cell-induced regeneration of the intervertebral disc [22], and further studies may be needed to better understand the potential effect of regenerative treatment.
This study has several limitations. First, functional outcomes were not evaluated. Second, patients were selected from a single institution. Therefore, due to the retrospective non-randomized nature of the study, a multicenter prospective randomized study is warranted to confirm whether other factors are associated with kyphotic deformity.