Introduction

The most common vertebral fractures occur in the thoracolumbar area, which is the transitional area of the spine (1-3). Thoracolumbar fractures occur most commonly in young adults (15 to 30 years of age) and may be associated with neurological deficits in 15 to 20% of patients (4,5). These fractures are proportionally on the increase, primarily because of the rising incidence of occupational and traffic accidents (1,2,4,5).

This upturn in the rate of thoracolumbar fractures has led to additional developments in surgical techniques and instrumentation technology. As a result, even patients with short life expectancies and poor quality of life related to a broad range of co-morbidities, who undergo surgery, seem to subsequently experience improved life expectancy and the ability to resume regular activities. Yet despite updated techniques, posterior, anterior, and combined surgical approaches have been used in the treatment of thoracolumbar fractures for the past five decades with several studies reporting excellent results (6-9).

The main surgical indications for thoracolumbar vertebral fractures are the associated presence of neurological deficit and vertebral instability (6,10-14). Nevertheless, a number of issues continue to hamper the process of determining the optimal management of thoracolumbar fractures. These include the lack of a widely recognized and validated thoracolumbar fracture classification, similar results obtained for some patient groups with either surgical or conservative treatment, and ambiguity about how best to understand and define the concept of spinal instability (3,15). Furthermore, the ideal parameters to use for measuring outcomes remain unclear.

In the present study, we aimed to measure a variety of clinical and radiological outcomes in a group of patients with thoracolumbar fractures who underwent surgery at a single center.

Materials and Methods

This retrospective study included 50 consecutive patients who underwent surgery for a thoracolumbar spine fractures between September 2000 and December 2011 at Ankara University Faculty of Medicine, Department of Orthopedics, by one of the two authors. We obtained approval from our institutional ethics committee for this study, and we conducted it in accordance with the principles of the Declaration of Helsinki. A written informed surgical consent was obtained from each patient.

We obtained clinical outcome responses from patients via the visual analogue scale (VAS) and the Oswestry disability index (ODI), which were both assessed preoperatively, postoperatively, and at the final visit. The 100-mm VAS was used to measure the intensity of pain associated with the fracture. The ODI was employed to measure the degree of disability associated with the fracture (16). If all 10 sections of the ODI are completed, the score can range from 0 to 50. Scores are calculated as a percentage (out of 50), and depending upon results, patients are described as minimally disabled (0% to 20%), moderately disabled (21% to 40%), severely disabled (41% to 60%), crippled (61% to 80%), or bed-bound / exaggerating symptoms (81% to 100%).

The Frankel scale, a 5-point severity scale, was used to determine the severity of spinal cord injury associated with the fracture (17). On this scale, spinal injuries are classified as complete (grade A), sensory only (grade B), motor useless (grade C), motor useful (grade D), or no neurological deficit (grade E). This was measured preoperatively, postoperatively, and at the final visit.

The sagittal index (SI), Local Kyphosis Angle (LKA), and anterior corpus height loss (ACHL) were measured using plain radiography preoperatively, postoperatively, and at the final follow-up visit. The SI is a measurement of the kyphotic vertebral segmental deformity corrected for the normal sagittal contour at the level of the deformed vertebral segment, and it was calculated as the angle between the posterior walls of the fractured vertebra and the intact vertebra immediately below it. The LKA is used to classify the sagittal plane deformity in the setting of traumatic thoracolumbar spine fractures, was measured using the Cobb method, and was defined as the angle formed between a line drawn parallel to the superior endplate of the intact vertebra one level above the fracture and a line drawn parallel to the inferior endplate of the intact vertebra one level below the fracture (18). The ACHL was calculated as the height of the fractured vertebra divided by the mean height of the intact vertebrae just above and below the fractured vertebra, and it was reported as a percentage.

Following the initial physical examination, patients underwent localized computed tomography (CT) and bilateral radiographs of any suspicious regions based on pain or tenderness. If examination revealed any neurological deficit, magnetic resonance imaging (MRI) was performed immediately. The vertebral fractures were classified based on the Thoracolumbar Injury Severity Scale and Score (TLISS) as well as the Denis classification system (5,10,19-21). The TLISS is based on three major injury characteristics: 1) the fracture morphology, 2) the integrity of the posterior ligamentous complex (PLC), and 3) the neurologic status of the patient. The Denis classification system divides the spine into three columns (and disruption of two or more columns results in instability): anterior column (anterior longitudinal ligament plus anterior half of vertebral body), middle column (posterior half of vertebral body plus posterior longitudinal ligament), and posterior column (pedicles, facet joints, supraspinous ligaments). The Denis system classifies fractures into four types - compression, burst, flexion-distraction, and fracture-dislocation - and differentiates each of these into five subtypes, A through E.

Patients most commonly remained hospitalized for three days after surgery. They typically underwent postoperative imaging one day after surgery. Postoperative VAS and ODI responses and Frankel scale measurements were generally obtained one week after surgery. Follow-up visits were routinely done at 1 week, 1 month, 3 months, 6 months, and 1 year after surgery. Final visit determination of VAS, ODI, and Frankel scale as well as imaging for calculation of SI, LKA, and ACHL, were done 1 year after surgery in most patients.

Statistical Analysis

Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS) Statistics for Windows, Version 17.0 (SPSS Inc., Chicago, IL, USA). Descriptive data are primarily presented as means. Friedman’s test and univariate logistic regression were implemented to assess and compare outcome results. Statistical significance was defined at the 5% (p<0.05) level.

Results

A total of 45 patients with vertebral fractures were admitted directly through our emergency department, whereas the remaining 5 patients were referred to our clinic from an external center. Of these 50 patients, 31 were female and 19 were male. The mean age was 46.5 (range: 16 to 76) years, and the mean follow-up was 96.5 (range: 6 to 183) months. All patients were evaluated and managed according to their trauma etiology and fracture level. The vertebral fractures were classified using both the TLISS system (Table 1) and the Denis classification (Table 2).

Of the 50 patients, 7 (14%) underwent anterior instrumentation and fusion, 41 (82%) underwent posterior instrumentation and fusion, and 2 (4%) underwent combined anterior and posterior instrumentation and fusion in the same session. Representative examples of preoperative and postoperative imaging of a patient receiving posterior instrumentation (Figure 1), postoperative imaging of a patient with a burst fracture (Figure 2), and postoperative imaging of a patient receiving anterior instrumentation (Figure 3) are provided.

The mean VAS score was 82 mm preoperatively, 60 postoperatively, and 13.5 mm at the final visit, and the mean ODI score was 65% preoperatively, 40% postoperatively, and 15% at the final visit. All of these improvements for both clinical outcome measures (including preoperative to postoperative, preoperative to final visit, and postoperative to final visit) were statistically significant (p<0.05). According to the Frankel scale, 42 (84%) patients had non neurological deficit preoperatively, 45 (90%) had no deficit postoperatively, and 46 (92%) had no deficit at final follow-up (Table 3).

Changes in mean SI values, LKA values, and ACHL percentages showed consistent trends from preoperative to postoperative or final visit, and from postoperative to final visit. All three radiological outcome measures demonstrated statistically significant improvements when preoperative values were compared to both postoperative and final visit values (all p<0.05) (Table 4). However, unlike clinical outcome measures (which progressively improved with time), all three radiological outcome measures actually demonstrated a loss of correction between the postoperative and final visits. Specifically, mean SI improved from 20° preoperatively to 14° postoperatively, before settling at 15° at the final visit. Mean LKA improved from 17° preoperatively to 9° postoperatively, before stabilizing at 13° at the final visit. Finally, mean ACHL improved from 45% preoperatively to 25% postoperatively, before settling at 28% at the final visit. However, the postoperative to final visit losses of correction for SI, LKA, and ACHL were not statistically significant.

Discussion

We retrospectively assessed the outcomes of thoracolumbar fracture surgery in 50 patients, using a variety of clinical and radiological measures, including VAS, ODI, Frankel scale, SI, LKA, and ACHL. Others have supported this approach by recommending that investigations seeking to quantify outcomes following spine trauma should employ a combination of existing surveys in a complementary fashion, and that these should include determinants of both bodily pain and work-related disability (22). Furthermore, whereas there are numerous radiological variables that can be evaluated following the surgical treatment of thoracolumbar fractures, SI, LKA, and ACHL appear to be used most commonly, and several studies using these parameters have shown varying short-term and long-term results (23-28).

For clinical outcome measures, we found that both the mean VAS and the ODI for all patients improved progressively and significantly from the preoperative visit to the postoperative visit to the final visit. Our results suggest that the severity of back pain was reported by patients to be minimal at the final visit. In addition, using the ODI definitions, patients reported that they had progressed from severely disabled prior to surgery to minimally disabled by their final visit.

For radiological outcome measures, we found that mean SI values, LKA values, and ACHL percentages all showed consistent trends from before surgery, to after surgery, and to the final visit. All three of these radiological outcome measures demonstrated statistically significant improvements when preoperative values were compared to both postoperative and final visit values. However, in contrast to clinical outcomes which continued to improve over time, all three radiological outcomes demonstrated declines, or losses of correction, between postoperative visits and final visits, though none of these changes were statistically significant. Thus, while trends in clinical and radiological outcome measures were similar between preoperative and postoperative visits, the trends in these outcome measures were dissimilar after that.

Others have also described a lack of correlation between radiological and clinical outcome measures. For example, Andress et al. (11) used the Hannover Spine score for the clinical evaluation of long-term results after surgery, and they did not find a significant correlation between improvements in LKA and clinical scores. They did report that clinical complaints were more frequent among patients with an LKA over 30 degrees; however, postoperative improvement of such a severe kyphosis angle is uncommon, which might explain the lack of correlation between improvements in LKA and clinical scores in their study. Similarly, Knop et al. (23) did not find a correlation between improvements in the Hannover Spine scores and any of the radiological outcome variables that they used. As a result, these authors suggested that radiological variables may not be useful for the long-term follow-up of patients with vertebral fractures.

Despite the findings that radiological outcome measures may not always correlate with clinical outcomes, these measures have still been utilized to assess the results in many studies. In 27 patients with short-segment thoracolumbar vertebral fractures, Wang et al. (29) found no significant correlation between the baseline or final severity of kyphosis and their pain scale, although 8 patients with an SI >15° showed a higher incidence of moderate to severe pain compared with the other 19 patients with an SI <15° (29). In another study, Liu et al. (30) undertook surgical treatment and follow-up of 18 patients using monosegmental transpedicular fixation plus posterior fusion. They used mean preoperative, postoperative, and latest follow-up SI values to demonstrate that their technique might provide the same or better fixation with the preservation of more motion segments among patients with thoracolumbar vertebral burst fractures with intact pedicles and facet joints accompanied by a PLC injury.

In their study of 50 patients undergoing short-segment vertebral posterior instrumentation, Andress et al. (11) reported a distinct improvement in SI values by restoring vertebral alignment but also a subsequent loss of correction of LKA during follow-up, reflecting alterations in the intervertebral disc space and the possibility of future degenerative disease. In their long-term study, Knop et al. (23) evaluated 62 patients who had surgery for thoracolumbar burst fractures. They reported a significant improvement in postoperative SI values, with no further alteration in these values during longer follow-up. They also found a mean loss of correction of LKA of 10°, despite also noting a significant improvement in the level of lordosis. Based on their study results, they concluded that LKA values tended to vary most in patients with a high preoperative ACHL percentage. In a related study, Toyone et al. (31) reported that the loss correction of LKA in the long-term was due to an unsupported anterior column. The authors recommended transpedicular intra corporeal hydroxyapatite grafting to address this, and they demonstrated that the loss of correction of LKA in patients in whom this technique was performed was significantly lower. The literature remains full of studies that have used radiological outcome measures to assess their results, suggesting the need to more definitively determine the value of these measures and to standardize how thoracolumbar fracture treatment results are assessed.

In this study, we were also able to assess the results of two different thoracolumbar fracture classification systems: the Denis classification and the TLISS. According to the Denis classification, the most common thoracolumbar fractures are burst fractures (5). Consistent with these data, 78% of our patients had burst fractures, followed by compression fractures in 16% and fracture-dislocations in 6%. Surgical indications for burst fractures include progressive neurological deficit, conservative treatment failure (new-onset neurological signs, increasing pain, unacceptable deformity), and fracture-dislocations (7). Of import, in patients with mechanically and neurologically unstable burst fractures, pulmonary and venous complications can be prevented, mobility can be maintained, pain can be relieved, spinal deformity can be minimized, decompression of neural components can be achieved, and disease progression can be halted through the use of surgery (7,32,33).

Similarly, based on TLISS fracture morphology characteristics, 78% of our patients had burst fractures, 16% had compression fractures, and 8% had fracture-dislocations. Also, based on TLISS, the vast majority of our patients (84%) had an intact neurologic status. However, in looking at the third component of TLISS, we noted that PLC injury was suspected or confirmed in 94% of our patients. This was important, because several studies have shown that the majority of significant thoracolumbar fractures present with PLC injuries, that MRI is most helpful to confirm the injury, and that surgical fixation is the optimal treatment in these cases (34).

Nevertheless, the optimal treatment for thoracolumbar spine fractures is still being debated. In a meta-analysis that included 275 articles pertaining to thoracolumbar burst fractures, Boerger et al. (35) reported only variable neurological improvement, irrespective of the technique, and they found no correlation between postoperative canal clearance and neurological improvement, suggesting that no surgical technique was superior in this scenario. However, most would agree that surgery is indicated in patients with a neurological deficit and/or fracture instability (6,10-14).

Many authors refer to the three-column concept described by Denis in assessing the stability of a spinal fracture (11,36,37). According to this concept, fractures that demonstrate damaged osteoligamentous (PLC) structures in the middle column on CT or MRI are unstable. Compared to stable fractures, unstable fractures are more often accompanied by a neurological deficit (36,37). The importance of the PLC to vertebral stability has become clearer in recent years, with a greater focus being placed on assessing damage to and stabilization of this structure, particularly in patients with compression fractures (38). Although most of our patients did not have a neurological deficit, nearly all of them had suspected or confirmed PLC injuries, suggesting that vertebral instability was likely and that surgical therapy with fusion was indicated.

With regards to surgical technique, the majority of patients in our study underwent surgery via a posterior approach. Compared to the anterior approach, the posterior approach has some advantages. It offers the ability to do surgery a safe distance away from the lungs and other visceral organs, resulting in lower morbidity and mortality rates (39). In addition, previous studies have shown that use of this approach also takes less time and is associated with a lower risk of bleeding (40). We looked at which approach was performed in the subset of our patients with thoracolumbar fractures who had associated neurologic deficits (according to TLISS) and showed an at least one grade improvement in their Frankel scale postoperatively. Of three patients with preoperative nerve root compression, two demonstrated Frankel scale improvement after surgery; of these, one underwent posterior instrumentation and fusion, while the other underwent anterior instrumentation and fusion. Only one of three patients with total cord compression demonstrated Frankel scale improvement after surgery, and this patient had posterior instrumentation and fusion using the posterior approach alone. The other two patients had combined anterior and posterior approaches.

Study Limitations

The implications of this study are limited by its retrospective design and the relatively small number of patients. In addition, given the divergent directions of clinical outcome and radiological outcome trends between the postoperative visits and final visits, measuring these outcome variables at various points between the postoperative and final visits may have provided additional valuable insights.

Conclusion

Trends in clinical and radiological outcomes after surgery for thoracolumbar vertebral fractures may differ. Although major progress has been made in the treatment of thoracolumbar vertebral fractures, the lack of standardized, verified clinical and radiological outcome measures continues to pose a challenge to the accurate assessment of the results of management.

Ethics

Ethics Committee Approval: Ankara University Faculty of Medicine.

Informed Consent: A written informed surgical consent was obtained from each patient.

Peer-review: Externally peer-reviewed.

Authorship Contributions

Surgical and Medical Practices: E.A., D.D., Concept: E.A., D.D., Design: E.A., D.D., Data Collection or Processing: E.A., Analysis or Interpretation: E.A., D.D., Literature Search: E.A., Writing: E.A.

Conflict of Interest: No conflict of interest was declared by the authors.

Financial Disclosure: The authors declared that this study received no financial support.