Although thoracolumbar burst fractures are common, the optimal therapeutic option still remains challenging [5, 6]. For neurologically intact patients without spinal instability, nonoperative treatment including short-term bed rest, brace or orthosis could achieve favorable results . The TLICS score proposed by Vacarro et al. has been widely used as a treatment algorithm for clinical decision making in thoracolumbar fractures . This novel classification considers 3 primary parameters including the fracture morphology, the neurologic status and the integrity status of posterior ligamentous complex (PLC) with a total of 10 points to determine stability and treatment option. According to this classification, nonoperative treatment is recommended for patients with a score less than 4. However, a quarter of initially conservatively treated patients might turn to surgery due to disabling pain . Moreover, the posttraumatic kyphotic deformity and back pain might progress after long term follow-up in nonoperatively treated patients despite adequate bracing . For thoracolumbar burst fractures with neurologic deficit, the TLICS score reached minimally 4 points and operative treatment is recommended. More recently, by combining the key benefits of TLICS and AO-Magerl classification, the AOSpine Classification Group proposed the AOSpine Thoracolumbar Spine Injury Classification System that simultaneously consider the morphological description of spinal column injuries, all major modes of failure and clinical features such as neurological status and treatment modifiers . And it may serve as a more valuable tools for communication, patient care, and research purposes in the future. In order to stabilize the spine, recover sagittal balance, decompress neural elements and obtain early patient mobilization, various surgical procedures including anterior, posterior, or combined approaches have been applied in the treatment of thoracolumbar burst fractures with neurologic deficit. However, there is still no consensus regarding the most suitable approach currently [5, 6].
Anterior approaches can achieve complete removal of the retropulsed bone and soft tissue fragments from the spinal canal under direct visualization without manipulation of the dural tube, as well as anterior reconstruction using plate or rod with bone graft . Theoretically, the anterior approach offers a satisfactory canal decompression and a better chance of neurologic improvement compared to other procedures . However, this approach is surgically challenging and more likely to have complications due to the adjacent chest and abdominal organs as well as major blood vessels . For example, Lin et al.  reported that patients in the anterior approach group had 5 times more complications but similar neurologic improvement comparing to the posterior approach group.
On the other hand, posterior pedicle screw–rod instrumentation could provide favorable stiffness and deformity correction due to its three-column fixation characteristics . Posterior approach with pedicle screw–rod instrumentation has been widely used for most thoracolumbar fractures nowadays [6, 8] since it can achieve favorable outcomes in terms of spine stabilization, kyphosis correction, postoperative neurologic improvement [9, 10, 26]. However, traditional open posterior operation requires massive paraspinal muscles stripping to expose the spinous process, lamina and facet, followed by short-segment (1 level above and below the injured level) or long-segment (2 levels above and below the injured level) internal fixation. The laminectomy is also performed at the injured level for canal decompression in patients with neurologic deficit. In this approach, massive paraspinal muscles stripping would cause ischemia, necrosis and denervation of the paraspinal muscle, resulting in atrophy and contractile properties loss of paraspinal muscles postoperatively. Denervation and dysfunction of the paraspinal muscles, as well as destruction of the posterior column stability, are believed to be associated with refractory postoperative back pain and disability . Recently, Li et al.  measured the cross-sectional area of the paraspinal muscle using MRI to compare the paraspinal muscle between the minimally invasive transforaminal lumbar interbody fusion (miTLIF) group and the traditional open TLIF group after the treatment of 1-segment lumbar disease. After 48 months follow up, patients in the traditional open TLIF group had significantly smaller cross-sectional area of the paraspinal muscle, complicated with worse back pain VAS scores and ODI scores, indicating the advantages of miTLIF in preventing paraspinal muscle atrophy, reducing postoperative back pain and improving postoperative life quality . Another disadvantage of the traditional posterior approach is that the canal decompression might be limited, especially when the canal encroachment is caused by repulsed bone fragments from injured vertebral bodies  and the posterior longitudinal ligament is likely to be injured, or the intra-canal fracture fragments are located in apterium of the posterior longitudinal ligament . Although some authors have reported no significant association between the extent of canal encroachment and neurological function [28, 29], a complete canal decompression theoretically offers a better chance for neurologic improvement and a more complete canal decompression with low risk of complications is worth trying for patients with neurologic deficit.
In this study, we presented a new approach via the Wiltse approach and the Kambin’s Triangle to achieve rigid fixation and direct decompression for the treatment of upper lumbar fracture with neurologic deficit. Compared to the traditional open posterior approach, this new approach has the following advantages: (1) The Wiltse approach enables pedicle screw implantation through natural intermuscular spatium without much muscle stripping and injury, thus preventing paraspinal muscle atrophy and reducing postoperative back pain. This was confirmed in our study that the pain level in the new approach group was significantly lower postoperatively and at 12 months follow-up. (2) The new approach can reduce intraoperative blood loss because major blood vessels are avoided through the natural intermuscular spatium. (3) The new approach is able to reach the ventral canal directly via the Kambin’s Triangle without the need for laminectomy, and the “L” shaped decompressor is applied to put pressure directly onto the intra-canal fracture fragments for reduction. Thus, the decompression becomes more efficient and complete without implicating the dura and neural elements. The average operation time was over 20 min shorter in the new approach surgery due to the avoidance of laminectomy. CT showed that canal encroachment was better relieved postoperatively and canal remodeling was more satisfied achieved at 12 months follow-up in the new approach group. Although neurologic recovery in the new approach group is similar to traditional posterior surgery group, the prior approach offers a more complete canal decompression and a theoretically better chance for neurologic improvement with shorter operation time and less blood loss. Moreover, if the intra-canal fracture fragments can’t be reduced by the decompressor, operators would get a feedback and laminectomy can therefore be performed as a remedy for decompression. The Kambin’s Triangle is a safe access to the disc and the spinal canal  and it has long been used in PELD and OLIF surgery as a minimally invasive technique for the treatment of degenerative lumbar diseases . In addition, partly removal of the anterior part of the superior articular process would not increase the risk of segmental instability , and can be performed in some cases to reduce nerve root stretching and facilitate decompression. (4) The present new approach can keep the integrity of posterior ligamentous complex (PLC) since it cause no further injury to the PLC, and the supraspinous ligament or interspinous ligament could be repaired by suture when ligament injury existed .
For severe unstable thoracolumbar burst fractures, posterior short-segment fixation alone without anterior support might result in implant failures. To determine certain fractures that would require supplemental anterior reconstruction, McCormack et al.  proposed the load sharing classification (LSC) determined by three components including comminution of the fractured body, apposition of fragments and kyphosis correction with a total of 9 points. According to McCormack et al., anterior reconstruction was necessary for patients with an LSC ≥ 7 because posterior short-segment fixation alone would lead to implant failures. Nevertheless, several techniques have been proposed to avoid complicated anterior surgeries and prevent implant failures in posterior surgeries for severe unstable thoracolumbar burst fractures (LSC ≥ 7) . The application of intermediate screws at the level of the fracture could increase the stiffness of the posterior short-segment construct and protect the anterior column during loading , avoiding the need for anterior reconstruction in the treatment of severe unstable thoracolumbar burst fractures (LSC ≥ 7) [33, 35]. Therefore, in the present new approach surgery, we preferred to insert bilateral or unilateral intermediate screws into the fractured vertebra depending on the integrity of the pedicles in order to increase fixation strength and facilitate reduction. A shorter intermediate screw (30-35 mm) was competent because pedicle contributes approximately 80% of the stiffness and 60% of the pullout strength at the screw-bone interface . Also, the Wiltse approach could provide broad operative field that was easy for implantation of pedicle screws and the prebent connecting rods. Our results showed a good reduction and kyphosis correction without implant failures in all the included patients through the posterior short-segment fixation with intermediate screws.
Limitations for current new approach should be noticed as well. Firstly, the present new approach is limited to only a portion of upper lumbar fractures. Thoracic spine fractures or patients with severe lamina fractures that have entrapped dural tissue or neural elements are not applicable. Secondly, its application relies on the special “L” shape tamps with various lengths and angles to adapt to different conditions according to each case. Thirdly, complications such as epidural veins bleeding and cerebrospinal fluid leakage during canal decompression are difficult to control in the present new approach, and open surgical technique may be needed in these conditions. Fourthly, as an innovative surgery, it has a learning curve and the beginners should be trained for a period of time, starting from some simple cases, such as burst fracture without large spinal canal compression. Lastly, more cases with longer follow-up are needed because current sample size and follow-up time are not adequate enough to draw a valid conclusion (Fig. 6).