Can the Interspinous Distance Predict Kyphosis in Conservative Treatment in Thoracolumbar Burst Fracture ? A Retrospective , Diagnostic Study

Study Design: Retrospective, diagnostic study. Objective: To verify if the interspinous distance is able to predict the risk for kyphotic collapse in thoracolumbar burst fractures treated conservatively without neurological deficit. Summary of Background Data: In patients with thoracolumbar burst fractures, the association between the amount of comminution, by using load-sharing classification (LSC), and kyphotic collapse is presented in the literature. However, LSC does not include the interspinous distance as an indirect sign to suggest biomechanical instability due to posterior ligamentous disruption in these patients in order to predict kyphotic collapse. Methods: We added the interspinous distance to the load-sharing classification (MLSC) in 50 consecutive patients with thoracolumbar burst fractures (according to Denis criteria) treated conservatively. Results: The LSC score was correlated to kyphotic collapse in the patients treated with TLSO (r = 0.312, p = 0.027; Spearman test; A = 0.668). The MLSC was similarly correlated to kyphotic collapse among TLSO-treated patients (r = 0.295, p = 0.038; Spearman test; A = 0.652). Conclusions: The interspinous distance did not contribute to the identification of worse radiographic outcomes, represented by the kyphotic collapse. This may suggest that the amount of comminution pointed out by the LSC is enough and more important than the interspinous opening in order to predict kyphotic collapse in thoracolumbar burst fractures. Possibly, the interspinous distance is much too heterogenous and multifactorial to be useful, since it reflects vertebral body height, preinjury anatomy, as well as posterior element disruption.


Introduction
The increasing incidence of spine injuries today inspires studies that help in the classification and treatment of patients with thoracolumbar fractures [1] [2].This region accounts for the vast majority of fractures in the spine.Of all types of fractures in this region, burst fractures are the most studied, due to its high frequency and high morbidity in multiple trauma patients, even in those without neurological deficits [2]- [4].
Holdsworth [5], in 1970, classified the thoracolumbar fractures based on the model of two columns, divided by the anterior longitudinal ligament, and they also described the burst fracture as a secondary injury to the compressive rupture of the vertebral body after an axial load.Two decades later, Denis [6] introduced the concept of the three parts of the column: anterior, middle and posterior columns.A burst fracture was classified as a major spinal injury affecting the anterior and middle columns [5]- [7].The Magerl classification [8] described the possibility, in these fractures, of posterior ligament injury associated with the retropulsed fragment into the spinal canal (B subtype).However, only the classification of McCormack et al. [7], known as the load sharing classification (LSC), suggests that the amount of comminution of the vertebral body can predict kyphotic collapse in these fractures [9]- [12].Some authors have studied the ability of McCormack scoring system to predict kyphosis collapse after conservative treatment in thoracolumbar burst fractures [12]- [14].In our view, there is a lack, in the scoring system proposed by McCormack, of variables addressing the increased interspinous distance as an indirect sign to suggest instability due to posterior ligamentous disruption in these patients.The most used method to investigate the posterior ligament insufficiency is the analysis of the opening of the spinous processes.Based on the anteriorposterior radiographic analysis of 200 normal subjects, Neumann et al. [15] defined the indirect values indicative of this failure that were accurate and reproducible among independent examiners.In clinical practice, most physicians believe that patients with increased interspinous distance have an indirect sign of biomechanical instability due to posterior spinal ligament insufficiency.We hope, therefore, to contribute to this discussion and present a complement of the LSC, adding the interspinous distance to the load-sharing classification (MLSC) to further studies.

Materials and Methods
the posterior spinal process distance, we used the method proposed by Neumann et al. [15] which quantifies the opening of the spinal processes in the posterior vertebral arch in radiographs in anterior-posterior (AP) view, taking into consideration the cranial end of the base of the spinous processes.The averages in adjacent levels were used as standards for normality.The distance was measured taking the cranial end of the "tear-drop" image of the spinous process in AP view (Figure 1).An upper limit of a normal difference in distance between the spinous processes at two adjacent levels was determined, in the evaluation proposed here, to be 7 -10 mm.Based on this value, the severity of the opening was graded as 1, 2 or 3, with 1 point for openings between 0 and 6 mm, 2 points for 7 to 9 mm and 3 points when 10 mm or over.These points were added to the LSC score, therefore the MLSC ranged from 4 to 12. Descriptive statistics of the data was made and to analyze the correlation between scores and Cobb angle of sagittal deformity, we used the Spearman correlation.We used ROC curves for patients with a loss equal to or greater than 10 degrees in order to define kyphotic collapse in these fractures during the follow-up.We considered the significance level of 5%, and used the software SPSS (Statistical Package for Social Sciences), version 13.0 in the statistical analysis.

Results
Fifty patients completed the criteria of this study..The LSC score was correlated to kyphosis deformity in the patients treated with TLSO (r = 0.312, p = 0.027; Spearman's test).The MLSC score was also correlated to kyphosis collapse among TLSO-treated patients (r = 0.295) 11 (22%) had a difference of 10˚ or higher (kyphotic collapse) compared to pre-treatment measurement.The ROC curve in Figure 2 illustrates the similarity between the two scoring systems to predict a kyphosis of 10˚ or higher (kyphotic collapse) in patients treated with brace (A = 0.668 and A = 0.652 for LSC and MLSC, respectively).

Discussion
Thoracolumbar burst fractures are found in patients suffering from multiple injuries from high-energy traumas [11].In these cases, detailed clinical and imaging exams are performed in search for parameters indicating injuries to the posterior capsular ligamentous complex [13] [14] [16].The most used currently is the posterior interspinous opening [7]- [9].It is interesting to note that, in practice, a lesion to the capsular ligamentous complex is suspected based on radiographic parameters according to the great majority of surgeons.Obviously, the ability to predict what fracture will deform would be of value to the clinician [18]- [21].
The signs on radiography and axial-CT that are characteristic of burst-type fractures described by Denis [6] were found by other authors [1] [11]- [17] [22]- [25] in injuries with flexion and distraction, since the mechanism of injury in these cases can aggregate axial forces that result in compression injury of the posterior column (sub type B of Margerl et al.) [8].In burst fractures, injuries in the posterior capsular ligamentous complex may be  suggested by the increase in the interspinous distance at the level of fracture in AP plain radiographs [18]- [20].
The scoring system proposed here (MLSC) adds the possibility to consider the increase in the interspinous distance, which is not included in the LSC classification.In patients without this interspinous opening, magnetic resonance (MRI), although more sensitive and specific to ascertain the posterior ligament injury, was not associated with worse clinical outcomes and radiographic findings in these patients [20].Another argument for not recommending routine MRI in these patients is the high cost of the test [18]- [20] [23] [24].This could be very important in regions where the access to the MRI would be difficult or even impossible.
Similar to our study, the correlation between the kyphotic collapse and the LSC in patients treated conservatively is pointed out by other authors [12].Although the MLSC has correlated with the kyphotic collapse in patients undergoing the conservative treatment, it did not increase the predictive power for kyphosis worsening compared with the LSC.This may suggest that the amount of comminution pointed out by the LSC is enough and more important than the interspinous opening in order to predict kyphotic collapse in thoracolumbar burst fractures.Possibly, the interspinous distance is much too heterogenous and multifactorial to be useful, since it reflects vertebral body height, preinjury anatomy, as well as posterior element disruption.At any rate, our results, even negative, must be looked at in future systematic review studies and MLSC considered in future prospective research.
We observed that the additional points referring to interspinous opening did not contribute to the identification of worst radiographic outcomes, represented by kyphotic collapse.We suggest studying the application of MLSC in patients with flexion-distracting (type B) and rotation (type C of Magerl) injuries.

Figure 2 .
Figure 2. ROC (receiver operator characteristic) curve for the scores and kyphotic collapse in patients treated with brace.Dashed line shows the load sharing classification (LSC), and dotted line shows the modified load sharing classification proposed in this study (MLSC)