Debridement and Bone Graft Fusion via the Lateral Extracavitary Approach Combined with Lateral and Posterior Screw-Rod Fixation in Treatment of Thoracic Spinal Tuberculosis: A Retrospective Study of 38 Cases


 BackgroundThoracic spinal tuberculosis is still common, and surgical treatment can rapidly relieve pain, correct deformity, reduce bone loss and prevent further damage to neurological function. We have practiced an efficient and safe surgical method.MethodsFrom January 2013 to April 2019, 38 patients with thoracic spinal TB were included in our study. Debridement and bone grafting were performed via the lateral extracavitary approach, combined with two different fixation method. Data from these cases were analyzed retrospectively.ResultsThe average surgical duration was 297.0±68.4 min, average intraoperative blood loss was 702.6±252.0 ml, postoperative hospital stay was 11.1±3.6 d, and C-reaction protein (CRP) and erythrocyte sedimentation rate (ESR) of all the patients decreased to normal levels at the last follow-up. The average Visual Analog Scale (VAS) score was 7.5±1.6 preoperatively and 0.6±0.8 at the last follow-up, which was significantly reduced compared with that before surgery. The average kyphosis correction was 6.3±4.7° and the angle loss was 1.4±1.6°. Neurological functions of all cases were significantly improved According to the American Spinal Injury Association (ASIA) classification. Solid fusion was observed in all cases at the last follow-up. One patient presented sinus tract formation at the incision site and the other two patients had rupture of the parietal pleura intraoperatively. No severe complications such as spinal cord injury or great vessel injury were found in all patients.ConclusionsDebridement and bone graft fusion via the posterolateral extracavitary approach combined with two fixation methods can both achieve high efficacy in the treatment of thoracic spinal TB. Lateral single screw-rod fixation is more suitable for patients with single-segment lower thoracic lesions and high stability, with less blood loss and shorter surgical duration. Posterior pedicle screw fixation has higher strength, and is more suitable for patients with multi-segment lesions, poor stability and complex conditions.


Background
Tuberculosis (TB) is more common in underdeveloped areas, but with the prevalence of immigration and human immunode ciency virus (HIV), TB infection also increases in Western world [1,2] . The spine is the most common site of extrapulmonary TB infection [3] . The spinal canal of thoracic spine is relatively narrow, which resulting in susceptibility to secondary neurological impairment once the abscess and kyphosis are formed here after TB infection [4] . Therefore, anti-TB treatment should be carried out so long as spinal TB is diagnosed. Most spinal TB can be cured by drug treatment [5,6] . While early surgical intervention is generally recommended in the cases of poor e cacy, severe vertebral bone destruction, deformity formation, spinal cord compression and neurological impairment. It has long been proposed that surgical treatment can rapidly relieve pain, correct deformity, reduce bone loss and prevent further damage to neurological function [7,8] . At present, the widely recognized treatment for spinal TB focuses on reconstruction of mechanical structure, thorough removal of lesions and improvement of neurological function, which is suitable for spinal TB at any segment. We advocate that thorough removal of lesions and prevention of recurrence should be the rst priority of three surgical focuses. Currently, there are many surgical approaches for thoracic TB. We selected the lateral extracavitary approach for debridement and bone grafting, through which the TB lesions in front of the thoracic spine can be well exposed, followed by fairly thorough debridement. As for the selection of xation methods, we chose lateral or posterior xation according to the number of involved segments, the site of lesions, spinal stability and vertebral bone condition, and achieved good results. After screening the patients treated in our group from January 2013 to April 2019, a total of 38 patients were included in this study. The clinical data of these patients will be analyzed below to verify the effectiveness and safety of this approach. For the convenience of analysis, some data will be displayed in different groups according to different xation methods.

Inclusion criteria
According to the following criteria, a total of 38 patients were selected from all TB cases treated in our group from January 2013 to April 2019. (1) The surgery was completed by the same surgeon, and it was the rst treatment; (2) diseased vertebrae were located from the superior margin of T1 to the inferior margin of T12, and the e cacy of conservative treatment was poor; (3) local structure was unstable, and estimated to aggravate; (4) patients had or were expected to have neurological impairment.

Basic information
All the patients were treated by debridement and bone grafting via the lateral extracavitary approach (Table 1). Among them, 18 patients received lateral single screw-rod xation, while the 20 patients underwent posterior screw-rod xation. There were 19 males and 19 females, aging from 23 to 84 years (average 51.0 ± 14.1 yeas). The follow-up time was 18-96 months, with an average of 50.8 ± 21.9 months. The involved vertebrae were distributed in T3-T12 (Fig. 1), with involved single segment in 25 patients and involved multiple segments in 13 patients. Preoperative ASIA classi cation revealed grade B in 2 patients, grade C in 3 patients, grade D in 11 patients, and grade E in 22 patients (Table 2).

Preoperative preparation
Preoperatively, standard anti-TB drug treatment (isoniazid 300 mg/day, rifampicin 450 mg/day, ethambutol 750 mg/day and pyrazinamide 750 mg/day) should be carried out for at least 2-4 weeks, and medical diseases should be actively treated to improve patients' tolerance to surgery.

Operative procedures
The patients with lateral xation were in the left or right lateral decubitus position according to the condition of lesions. A paraspinous incision was made on the back, with the trapezius/latissimus dorsi separated layer by layer. Through the intermuscular space of the erector spinalis, the ribs connected with the diseased vertebrae and one adjacent upper and lower ribs were exposed. The spaces between the ribs and the anterior parietal pleura were separated carefully to free the ribs. At about 10 cm of the distal end, the ribs were cut off and removed completely. The adjacent ribs were removed in the same way (the number of removed ribs depended on the need of debridement). Afterwards, the intercostal vessels and nerves were ligated and severed. The transverse processes of the diseased and adjacent vertebrae were cut off using an osteotome, so the anterolateral surface of thoracic vertebrae could be reached (Fig. 2). Pus, diseased intervertebral disc and bone were removed completely. If spinal canal stenosis was severe preoperatively, lateral spinal canal decompression could be performed by opening the spinal canal along the intercostal nerves. After thorough debridement, one pedicle screw was inserted into the lateral upper and lower vertebral bodies, respectively, crossing the lesion. After washing, a massive iliac bone, rib or titanium mesh was implanted. Subsequently, the titanium rod was connected, bleeding was stopped thoroughly, 2 g streptomycin was placed locally, and the incision was closed after placing a drainage tube.
The patients with posterior xation were in the prone position rstly. A posterior-midline incision was laterally made on the back with the diseased vertebrae as the center for exposure till the vertebral plate. The pedicle screw was inserted routinely, and the screw rod on the opposite side of the approach was locked. The incision was closed temporarily, and the patients were changed to the lateral decubitus position. Both ends of the original incision were extended obliquely upward for layer-by-layer exposure. In the same way, the exposure was performed till the anterolateral side thoracic vertebrae, which was followed by careful debridement. If there was spinal canal stenosis, lateral spinal canal decompression could also be carried out. After repairing the bone defect and washing, an autogenous bone or titanium mesh was implanted and connected with the titanium rod on the approach side, which was locked nally. Afterwards, bleeding was stopped thoroughly, 2 g streptomycin was placed locally, and the incision was closed after placing a drainage tube.

Postoperative management
The intraoperatively pathological tissues were sent for pathological examination. In this study, all patients were con rmed as TB by pathological examination. After surgery, all patients received anti-TB drug treatment, routine prevention of general bacterial infection, positive correction of anemia, and body uid balance on the next day. The character of drainage uid in the drainage tube were observed, and the drainage tube was withdrawn when there was no extraction. After the general condition was stable and the incision was normal, the patients were discharged. All the patients were asked to lie in bed for 8 to 12 weeks, and then out-of-bed activities with brace. CRP, ESR and computerized tomography (CT) or X-ray were reviewed at 3, 9 ,12 and 24 months after surgery, respectively. Anti-TB drug treatment lasted at least 18 months after surgery.

Evaluation
Surgical duration, intraoperative blood loss and postoperative hospital stay (d) were used to assess the surgical process. Preoperative and postoperative VAS score, and preoperative ESR and CRP were used to assess the postoperative recovery, and ASIA classi cation to assess neurological function. X-ray and CT were used to assess bone graft fusion and the correction of spinal deformity. The kyphosis angle was measured on the lateral position of the X-ray images: a horizontal line was made at the superior margin of the upper vertebral body and the inferior margin of the lower vertebral body involved in kyphosis formation, and the angle between the two lines was the kyphosis angle. The kyphosis angle was followed up to assess the correction. Additionally, the complications of all patients were recorded.

Statistical analysis
The statistical analysis in this study was performed by SPSS 24.0 statistical software (IBM Corp., Armonk, NY, USA). A paired t-test was applied to compare the changes in indices preoperatively, postoperatively, and during follow-up. A P value of < 0.05 was considered statistically signi cant.

Perioperative data
The average surgical duration was 297.0 ± 68.4 min, average intraoperative blood loss was 702.6 ± 252.0 ml, postoperative hospital stay was 11.1 ± 3.6 d. The average preoperative CRP and ESR were 18.32 ± 24.1 mol/L and 30.7 ± 23.5 mm/h, respectively. At the last follow-up, CRP and ESR of all patients decreased to normal levels, 1.1 ± 1.2 mol/L and 8.2 ± 4.4 mm/h, respectively. The average VAS score was 7.5 ± 1.6 preoperatively and 0.6 ± 0.8 at the last follow-up, which was signi cantly reduced compared with that before surgery (P < 0.05) ( Table 1).

Complications and neurological function
Two months after surgery, one patient undergoing posterior xation and debridement via the lateral approach showed sinus tract formation at the incision site, which was con rmed as recurrence of the original lesion and cured after debridement. Nine months after second surgery, two screws at the lower end were withdrawn. While CT con rmed that boney fusion was formed at the bone graft site. In the other patients, the incisions healed at the rst stage, and no internal xture fracture or withdrawal was found. Two patients presented rupture of the parietal pleura during surgery, which were repaired intraoperaively in time. No intraoperative dural tear and large vessel injury, or spinal cord injury, cerebrospinal uid leakage and intractable intercostal neuralgia occurred in any patients. As for neurological function ,a total of 16 patients were lower than grade E before surgery, and at the last follow-up, one patient was upgraded from Grade B to Grade D, while the rest of the patients were upgraded to E( Table 2).

Follow-up of imaging data
The kyphosis angle was 16.8 ± 7.9° before surgery, 11.6 ± 5.9° after surgery, and 0° in the last follow-up.

Comparison between different internal xation groups
In addition, according to different internal xation methods, comparison was performed between groups. The main differences are as follows, as shown in all tables. Firstly, in the posterior xation group, a total of 12 patients had multi-segmental involvement, while in the lateral xation group, only one patients showed multi-segmental involvement. Secondly, the involved segments were distributed throughout the thoracic spine in the posterior xation group, while only below T6 in the lateral xation group.
Preoperative kyphosis angle of the posterior xation group was 18.8 ± 6.9, which was larger than that of the lateral xation group (14.5 ± 8.5). The kyphosis correction was 6.5 ± 4.6° in the posterior xation group and 6.1 ± 4.8° in the lateral xation group. Moreover, there were four patients in the posterior xation group whose preoperative ASIA grade was lower than D, while one patients in the lateral xation group. Not surprisingly, the surgical duration was signi cantly shorter and blood loss was signi cantly less in the lateral xation group than those in the posterior xation group.

Discussion
Spinal TB usually involves the thoracic vertebrae, and bone destruction often occurs in the anterior and middle column. Therefore, the anterior spine must be exposed for thorough debridement. The lateral extracavitary approach is widely used in the surgery for ventral lesions of the thoracic spine, including fractures, tumors, infections, etc., with high e cacy [9] . In the treatment of thoracic TB via this approach according to our practice, the following key points should be paid attention to: the approach should be selected according to preoperative imaging data, and the patients should be in the lateral decubitus position, with the approach upwards. For the lower thoracic vertebrae, the transverse part of the incision should be located about 5 cm laterally to the posterior midline (Fig. 3). If posterior xation is selected or the lesion is located in the upper thoracic vertebrae(above T6), the transverse part of the incision should be biased to the center to facilitate nailing in the prone position. Then the incision was closed temporarily, and the patients were changed to the lateral decubitus position. Both ends of the original incision were extended obliquely upward for layer-by-layer exposure. After the latissimus dorsi or trapezius are cut open layer by layer and the deep erector spinalis is exposed, separation should be performed with erector spinalis reserved, till the ribs. The ribs need to be dissected carefully from the anterior periosteum using a periosteum stripper, then cut them off respectively and remove completely, during which the parietal pleura should not be damaged. In addition, the transverse process should be cut off simultaneously. Once parietal pleura is fractured, it should be repaired immediately. After ligation of intercostal blood vessels and nerves and disconnection, dissection should be performed carefully along the surface of the thoracic vertebrae, then the anterior column of thoracic spine can be reached through the parietal extrapleural space (Fig. 3). If spinal canal decompression is necessary, the anterior lesions should be cleared roughly for preliminary decompression,which can release the pressure in front of spinal cord and avoid injury of it during opening the spinal canal. Subsequently, the partial vertebral plate and pedicle, and the tissue protruding backwards in the anterior spinal canal should be resected for su cient decompression (Fig. 4,Fig. 2).
At present, there are multiple surgical approaches for thoracic TB, including anterior approach, posterior approach ,and combined approach, which have their own advantages. The anterior approach needs to enter through the thoracic/abdominal cavity, which was rstly reported by Hodgson et al. [10] . It can be used for debridement, precise bone grafting and spinal reconstruction in one stage, with high e cacy. The combined anterior and posterior approach is also widely used, which can achieve 360° spinal reconstruction. Based on thorough debridement and nerve decompression, it has better orthopedic ability and stability. Hirakawa, Liu et al. [11,12] reported that the treatment of thoracic TB via this approach achieved good results. Wang, Guzey, et al. [13,14] reported that the treatment of thoracic TB via the posterior only approach achieved high e cacy. Debridement, bone grafting and internal xation were completed in one stage, and the debridement was thorough and the xation was reliable. They also compared the advantages and disadvantages of the three surgical approaches, and concluded that the posterior approach alone should be the rst choice for the treatment of thoracic TB. In addition, Yin [15] also reported that the lateral extracavitary approach for the treatment of thoracic TB involving multiple segments achieved high e cacy, which is in line with the approach in our study. Combined with our practical experience and data, it is concluded that the lateral extracavitary approach has the following advantages: (1) it provides wide eld of vision, thorough debridement and precise bone grafting. A window can be made by resecting the ribs the transverse processes connected with the diseased vertebrae. If multiple segments are affected, several adjacent ribs can be resected to enlarge the window. Through this window, the anterolateral side of thoracic spine can be reached. Under direct vision, the abscess and necrotic tissues in front of the every affected vertebra can be removed (Fig. 5). Moreover, in the lateral decubitus position, the lesion on the opposite side of the thoracic vertebra can be reached after gradual debridement via this approach, so that debridement is more thorough. After debridement, the amputated rib or ipsilateral iliac bone can be used for structural bone-grafting, or it can be crushed and lled into titanium cage, and then implanted into bone groove. (2) Whether the spinal canal should be opened can be determined according to the demand (Fig. 2). When the anterior lesion is cleared preliminary, if the spinal canal stenosis is con rmed by preoperative imaging data, the spinal canal can be opened laterally at lesion space for decompression. In some cases, the anterior vertebral bone destruction is severe, but the spinal canal is not involed. In some cases, there is no need for decompression, but only debridement (Fig. 2,Fig. 3), the resection range only involves the transverse processes, ribs and Pathological bone. The integrity of the spinal canal can be preserved, meanwhile the remaining support of the spine can be retained as much as possible, then the risk of spinal cord injury can be reduced. The obvious disadvantages of this approach include: large trauma, and need for disconnection of intercostal nerves and blood vessels due to rib resection, resulting in postoperative chest wall paresthesia in some patients. Furthermore, if posterior xation is selected, the position should be changed during surgery, which also increases the surgical duration and the risk of infection.
The orthopedic ability and stability of posterior xation are signi cantly better than those of lateral single screw-rod xation [16] . While, our data showed that lateral single screw-rod xation could achieve the same orthopedic effect as posterior xation, and the loss of kyphosis angle was not obvious. The cause is that the choice of internal xation method was not random during our diagnosis and treatment. In the cases of single screw-rod xation, the lesions were relatively simple and the stability was not bad. In contrast, posterior xation was selected for the patients with severer conditions, more bone destruction, more complex lesions and poorer spinal stability after debridement. Additionally, the affected segments of the 18 patients with lateral single screw-rod xation in our study were all below T6. Because lateral screw-rod xation is not suitable for the upper thoracic spine (T1-T5). The lateral side of the upper thoracic spine is covered by the scapula, the curvature of the ribs is greater, and the pleura is tenser, which make lateral screw placement more inconvenient. Therefore, for lesions in the upper thoracic spine (above T6), we chose posterior xation. Combined with practice and data, we preliminarily explored the indications of the two xation methods. Based on debridement via the posterolateral approach, posterior xation should be selected for patients with multi-segment involvement, complex lesions, severe deformity and poor stability (Fig. 6). In the case of single-segment lesions of the lower thoracic spine, good stability and not serious deformity, lateral single screw-rod xation can be selected (Fig. 7). Their own characteristics are as follows: Posterior xation has good holding force and supporting capacity, with high strength, high stability and wider application range. Lateral single screw-rod xation can be completed with debridement and bone grafting in one stage. It is characterized by relatively small trauma, short surgical duration, less blood loss and simple operative procedure,but features weak strength. In addition, Mycobacterium tuberculosis will not form a bio lm on the surface of the internal xation implant [17] , so it is not improper to directly implant the screw rod or titanium cage into the lesion.
Finally, any surgical method should be selected based on the patients' condition combined with the surgeons' expertise, so as to achieve individualized treatment. Anti-TB drugs are the basis of the treatment of TB. While paying attention to surgical treatment, drug treatment should be given enough attention. The follow-up time of this study is relatively short, and the complications need more long-term observation.

Conclusions
Debridement and bone graft fusion via the posterolateral extracavitary approach combined with two xation methods can both achieve high e cacy in the treatment of thoracic TB. Lateral single screw-rod xation is more suitable for patients with single-segment lower thoracic lesions and high stability, with less blood loss and shorter surgical duration. Posterior pedicle screw xation has higher strength, and is more suitable for patients with multi-segment lesions, poor stability and complex conditions.

Consent for publication
Informed consent was obtained from all individual participants included in the study.
Availability of data and materials One case was xed with lateral single screw rod without spinal decompression. (a) The patient is in the right decubitus position with the incision at the solid line. (b) The two white arrows show the ribs attached to the diseased vertebra. The lateral side of the thoracic vertebra is exposed when the ribs and transverse processes are removed while the anterior tissue is pushed forward. (c) The creamy uid is pus exudating from the intervertebral space. (d)After debridement, titanium cage was placed at the bone defect, and 2 pedicle nails was implanted at the side of two vertebral bodies.