Long-term results after thoracoscopic anterior spondylodesis with or without posterior stabilization of unstable incomplete burst fractures of the thoracolumbar junction: a prospective cohort study

ODI evaluation reported a mean spine-related impairment of 11.7% (Table 3). An analysis of the isolated anterior stabilization (anterior_only) and combined anterior and posterior stabilization (combined_AP) subgroups demonstrated comparable long-term results. Thirteen patients had minimal limitations with six patients reporting no spine-related limitations in sports and activities of daily living. One patient reported moderate limitations in the ODI questionnaire (24%) and complained about lasting donor site pain. Overall, three patients complained about lasting donor site-related symptoms. One patient complained about persisting pain in the paravertebral muscles despite posterior implant removal (ODI 10%, PCS 51.9 pts, MCS 54.5 pts). One patient was severely limited (ODI 46%) and had to change work due to back pain (VAS spine 35.2) and donor site symptoms; this patient reported severe limitations with an ODI of 46%. Notably, this individual had suffered a polytrauma (Injury Severity Score of 19) at the index presentation, which included extensive abdominal injuries (splenic rupture, small-intestine rupture) and fracture of the upper extremities (fracture of distal radius AO type 23C3.3). The postoperative symptoms were not exclusive to the patient’s back issues but also related to abdominal discomfort and upper extremity symptoms.

No significant difference was found in relation to disc height or loss of disc height in the adjacent levels between the different operative strategies. Two patients displayed the formation of new osteophytes (> 3 mm in size) in the adjacent segments, but there was no association with anterior or posterior spondylolisthesis. No relevant loss of disc height (> 20% compared to a cranial reference disc) was found in the cohort. Furthermore, no significant correlation was found between the loss of disc height and increase of regional malalignment.

In this investigation, ODI evaluation reported a mean impairment of 11.7%, which equates to minimal disability at the latest follow-up of 12 or more years. Unfortunately, these results cannot be compared over time as ODI was not evaluated at the time of patient inclusion and at the 6-year follow-up. However, it is encouraging that these scores mimic the values reported by Scholz et al. [7] in a short-term follow-up (2 years) of patients treated with a combined anterior and posterior stabilization using the MIAS technique (ODI score of 13.3%). These positive results were reinforced by an analysis of the SF 36 domains in our study demonstrating no differences between the 6- and 12-year follow-up as well as in comparison to a healthy control group of the same age.

Notwithstanding, non-operative treatment is a generally accepted treatment alternative for incomplete burst fractures of the thoracolumbar junction without neurologic deficits [3]. Thereby, Wood et al. [3] presented favorable long-term clinical results for non-operative treatment of thoracolumbar A3/A4 fractures. They presented a median ODI score of 4% after a follow-up of 18.6 years for those treated non-operatively whereas the cohort treated operatively had a median ODI score of 40%. However, despite the better clinical outcome following non-operative treatment, a relevant bisegmental sagittal loss of reduction was observed in this collective with an average Cobb angle of 19° (10–29°). This might lead to a decompensated sagittal alignment in the future with associated symptoms. Interestingly, Wood et al. [3] reported on far higher rates of ASD of 70% in the non-operative group and 64% in the operative-treated group compared to 15% in our patient population.

Two further studies reported on long-term outcome after non-operative treatment of thoracolumbar burst fractures. Weinstein et al. [19] re-evaluated 42 patients with thoracolumbar burst fractures at a mean follow-up of 20 years. They found an average Oswestry Low Back Pain Score of 3.5 in a heterogeneous cohort of thoracolumbar and lumbar fractures. Taking into account that we used a VAS spine score with an inverted scale, these findings are slightly worse than the results of our cohort. Further, Moller et al. [20] followed 27 patients with a 23–41 year follow-up. Twenty-one patients (78%) showed a mean ODI score of 8% demonstrating a slightly better clinical outcome compared to our cohort. However, Moller et al. reported on a heterogeneous collective with a wide range of fracture location and fracture morphology with a follow-up rate of only 51%. Although the average bisegmental kyphosis was reported to be > 20°, there was no correlation between alignment and clinical outcome. In summary, a clearly superior regional long-term alignment is visible in our patient collective treated surgically using MIAS compared to a non-operative therapy. However, we could not proof a clinical benefit of our surgical strategy. This might be caused by our inclusion criteria of unstable fractures with a load sharing classification of 6 and higher [14]. Further matched-paired or randomized control trials need to evaluate this important question in more detail.

Additionally, the proportion of patients complaining about approach-related symptoms after iliac crest harvesting remained high which might explain the slightly inferior clinical results. Three patients complained about pain or hyposensitivity at 12 year follow-up. At 6-year follow-up, 71% of all patients were suffering from donor site pain [21]. These findings are consistent with the findings of Wippermann et al. [22] who reported an overall complication rate of 19.6% after iliac crest harvesting. Based on these results, the authors do not implant iliac crest grafts any longer. We changed our therapy regime to trabecular metal cages for monosegmental and distractable titanium cages for bisegmental vertebral body replacement. At 12-year follow-up, the morbidity of the anterior thoracoscopic approach was still negligible with only one patient complaining about mild regional hyposensitivity.

In our cohort, radiographic examination revealed a moderate bisegmental increase of regional malalignment of 9° at the latest follow-up. Progression of regional malalignment after complete bony healing is surprising at first sight. We believe this can be explained by a continuous loss of vertebral disc height in the adjacent motion segments on the one hand and by an ongoing remodeling process of the bone graft in patients treated with monosegmental spondylodesis on the other hand. Low to moderate cage subsidence might explain this situation in patients treated bisegmentally. Interestingly, there was a higher increase of regional malalignment in the anterior_only group compared to the combined_AP stabilized group. This might be caused by the improved initial reduction after combined_AP treatment. However, the differences were not found to be significant, although this may be due to the small sample size.

Recently, Smits et al. [10] reported promising results in 46 patients with thoracolumbar fractures treated by MIAS at a median follow-up of 49 months. Thereby, the overall bisegmental loss of reduction in this cohort was comparable to our results with a mean of 6.8°. However, Smits et al. did not investigate the effects of the MIAS technique on spinal alignment and included a more heterogeneous group of patients with a diverse variety of fracture types, neurological status, fracture locations, and time to follow-up.

In contrast, Scholz et al. [7] found a mean loss of reduction of only 1.9° in the combined_AP treated subgroup. This lower loss of reduction might be caused by the far shorter follow-up of only 2 years. Similarly, in the previous investigation of our cohort, Spiegl et al. [9] reported a loss of reduction of 4.0° at 18 months follow-up but reported a relevant progression in the further course after removal of the posterior implants. Korovessis et al. [5] reported a mean bisegmental loss of reduction of 2° over 46 months in a combined_AP stabilized cohort and a mean loss of 5° over 48 months in a posterior only stabilized subgroup with mid-lumbar A3 fractures. Hereby, the reduced loss of reduction can be explained by the fracture location at the lordotic mid-lumbar spine.

Additionally, a higher reduction loss can be caused by the implantation of autografts in the majority of our patients. Lack of intrinsic stability of iliac crest bone grafts and loss of height due to late onset bony healing are possible reasons for a worsened radiologic outcome [21]. In combination with frequent donor site symptoms, this led to a replacement of autografts by cages in our department.

ASD is an acknowledged problem after segmental spinal fusion in treatment of degenerative disc disease [23, 24]; increased movement and mechanical loading on the adjacent discs are potentially causative factors. After definition of radiologic signs of ASD, Tsuji et al. [16] reported 21 cases of ASD in a cohort of 71 patients who underwent PLIF-operation after 5 years. Although the underlying conditions are not directly transferable, ASD might be a long-term problem after trauma-related spinal fusions, too. As previously reported, Wood et al. [3] reported on rates of ASD of 64% in the intervention group and 70% in the non-operative group in trauma patients. However, Wood et al. [3] did not define limit values for radiological signs of ASD in detail. In our study, we found new onset of osteophytes in the adjacent segments in two out of 12 patients. Other signs of degeneration like disc space narrowing or listhesis were not observable. One possible reason for this relatively low rate of ASD is the younger average age of our patients collective at final follow-up (52 years vs 62 years Wood et al.). However, the most likely parameter for ASD progression is the high regional malalignment considering the high kyphotic Cobb angle of 19° in the latest follow-up.

Overall, no statistically significant correlations between alignment parameters and clinical outcome scores could be found. These results are in accordance to other studies which state that patients do not necessarily present with a clinical deterioration in cases with mild to moderate kyphotic malalignment [5, 19]. However, Gertzbein et al. [25] found significantly higher rates of back pain in patients with kyphotic malalignment of more than 30°. In our cohort, the mean loss of reduction was 9° which might be too small to affect compensation mechanisms, and the maximum kyphosis of 22° was also lower than the critical amount of 30° [25]. All patients showed balanced sagittal alignment parameters with a perpendicular spine profile. Overall, the interpretation of effects remains difficult due to a wide range of data and variations between individuals [26].

The major limitation of this study is the low number of included patients. Thus, the study might be under-powered to detect differences between the anterior only and combined anterior and posterior therapy strategy. However, the main purpose of this study was to evaluate long-term clinical und radiological results of MIAS in a highly specific patient group (affected vertebral bodies of the thoracolumbar junction only, acute A3 fracture, no osteoporosis). The authors were particularly interested in evaluating approach-related symptoms and analyzing the long-term effect for ASD progression and sagittal alignment.

Additionally no randomization has been performed leading to an inhomogeneous distribution of fracture types and grades of instability between the two subgroups in the index study. Hereby, it would be interesting to compare the anterior only strategy with a combined anterior and posterior treatment in a bigger patient population.

Furthermore, clinical investigation and full spine radiographs were not available in all cases at the latest follow-up. Besides, sagittal alignment and the results of the ODI evaluation cannot be put in a chronological correlation as both had not been used in the preceding investigation. However, both the ODI score and full spine radiographs were not commonly evaluated in 2002. Besides, two different operative strategies (anterior_only vs combined_AP) were used and, in the meantime, the operation technique for MIAS using iliac crest bone crafts had been widely replaced with titanium cage implementation.

However, the strengths of this study are the prospective data acquisition and the high comparability of the analyzed fractures as only A3 fractures of the thoracolumbar junction were included. Loss of follow-up was excellent with less than 20% over 12 years.