Posterior short-segment fixation in thoracolumbar unstable burst fractures – Transpedicular grafting or six-screw construct?

Posterior short-segment fixation in thoracolumbar unstable burst fractures – Transpedicular grafting or six-screw construct?

Clinical Neurology and Neurosurgery 153 (2017) 56–63 Contents lists available at ScienceDirect Clinical Neurology and Neurosurgery journal homepage:...

1MB Sizes 0 Downloads 108 Views

Clinical Neurology and Neurosurgery 153 (2017) 56–63

Contents lists available at ScienceDirect

Clinical Neurology and Neurosurgery journal homepage: www.elsevier.com/locate/clineuro

Posterior short-segment fixation in thoracolumbar unstable burst fractures – Transpedicular grafting or six-screw construct? Jen-Chung Liao ∗ , Kuo-Fon Fan Department of Orthopedics Surgery, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan

a r t i c l e

i n f o

Article history: Received 24 September 2016 Received in revised form 19 December 2016 Accepted 20 December 2016 Available online 21 December 2016 Keywords: Thoracolumbar burst fracture Posterior short-segment fixation Six-screw construct Injectable calcium sulfate/phosphate cement Load-sharing classification

a b s t r a c t Objectives: Early implant failure and donor-site complication remain a concern in patients with thoracolumbar burst fracture underwent one-above and-below short-segment posterior pedicle screw fixation with fusion. Our aim was to evaluate the results of short-segment pedicle instrumentation enforced by two augmenting screws or injectable artificial bone cement in the fractured vertebra, and compare the differences between these two Patients and methods: We conducted a retrospective clinical and radiographic study. Twenty-seven patients were treated with a six-screw construct (group 1), and twenty-nine patients underwenta fourscrew construct and fractured vertebra augmentation by injectable calcium sulfate/phosphate cement (group 2). Posterior or posterolateral fusions were not performed in both groups. The severity of the fractured vertebra was evaluated by the load-sharing classification (LSC). Local kyphosis and anterior body height of the fractured vertebra were measured and were follow-up at least 2 years. Any implant failure or loss of correction >10◦ degrees at the final was defined as failure of surgery. Patients’ clinical results were assessed by the Denis scale. Results: Blood loss and operation time were less in group 1 (126.2 ± 9.7 vs. 267.6 ± 126.1 ml, p < 0.001 and 141.2 ± 48.7 vs. 189.8 ± 16.4 min, p < 0.001). Immediately after surgery, group 2 had a better local kyphosis angle (3.7 ± 5.3 vs.6.0 ± 4.1◦ , p = 0.047) and acquired more anterior body height (94.9% ± 7.6% vs. 84.9% ± 10.0%, p < 0.001). Both groups had similar clinical results (pain score: 1.5 ± 0.8vs. 1.4 ± 0.6, p = 0.706; work score: 1.7 ± 0.9 vs. 1.6 ± 1.0, p = 0.854). Group 1 had 3 cases of surgery failure; group 2 had 8 cases of implant failure (p = 0.121). The average LSC score of these 11 patients with surgical failure was 7.2. Conclusion: Thesix-screw construct had the advantage of shorter operating time, less blood loss, and lower failure rate. For those patients with anLSC score 7, posterior short-segment instrumentation should be used cautiously. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Burst fracture approximately accounts for 20% of thoracolumbar fractures and occurs due to an axial loading force that results in failure to support the anterior and middle column [1,2]. Proper management of thoracolumbar burst fractures remains controversial and includes nonsurgical and surgical treatment. For those neurologically intact patients with minor deformity, nonoperative treatment of short-term bed rest followed by a molded thoracolumbar orthosis can achieve excellent results [3,4]. Surgery is usually

∗ Corresponding author at: Department of Orthopedic Surgery, Chang Gung Memorial Hospital, Chang Gung University, No. 5, Fu-Shin Street, Kweishian, Taoyuan 333, Taiwan. E-mail address: [email protected] (J.-C. Liao). http://dx.doi.org/10.1016/j.clineuro.2016.12.011 0303-8467/© 2016 Elsevier B.V. All rights reserved.

indicated for a patient suffering from severe deformity, for severe pain limiting the patient’s activity level, and/or neurologic deficit. The type of surgery includes anterior surgery, posterior surgery, a combination of anterior and posterior surgery, and minimally invasive surgery [5,6]. One-above and one-below posterior short-segment instrumentation with fusion has been widely used for unstable thoracolumbar burst fracture for the past 3 decades [7]. Pedicular instrumentation enables kyphosis correction, indirect reduction of canal encroachment, and early mobilization. However, this method has been reported to have a high rate of implant failure and early loss of reduction because of loss of anterior support [8]. Furthermore, donor-site complications, including infection and hematoma formation are a major concern if the bone graft is harvested autogenously [9].

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63 Table 1 Inclusion criteria of the study. 1. 2. 3. 4. 5.

6.

57

Table 2 Denis Pain Scale and Work Scale. Fracture type: A3 burst fracture (AO Magerl classification) Fracture level: T11 to L3 Surgical method: short-segment posterior instrumentation Fusion method: no posterior or posterolateral fusion applied Fractured vertebra augmentation: two additional screws or injectable calcium sulfate cement Follow-up period: at least 2 years

AO = Arbeitsgemeinschaft für Osteosynthesefragen.

Liao et al. and Korovessis et al. have demonstrated that injectable calcium sulfate cement or injectable calcium phosphate cement used as a transpedicular grafting material in thoracolumbar fracture could obtain clinical and radiographic results equal to autogenous cancellous bone graft, but donor site complications could be prevented [10,11]. A recent biomechanical study of an unstable thoracolumbar burst fracture found that 2 additional screws in the fracture vertebra (a six-screw construct) could increase the stiffness of the implant and reduce stress on each pedicle screw, compared to the four-screw construct [12]. In a clinical study, Wang et al. showed that posterior short-segment fixation with an additional pedicle screw at the fracture level in thoracolumbar burst fracture, but without posterior fusion, provided two-year follow-up results similar to those with fusion [13]. From the above literature review, the use of posterior shortsegment instrumentation without fusion in treating thoracolumbar burst fracture seems to be a trend. To our knowledge, there are no studies comparing the difference between these 2 types of posterior short-segment instrumentation without fusion for thoracolumbar burst fractures. This study was to evaluate the efficacy of these 2 methods of posterior short-segment fixation without posterior fusion in the treatment of thoracolumbar burst fractures. We hypothesized that two additional pedicle screws at the fractured vertebra could augment construction of instrumentation and the spine could be maintained when the anterior and middle column achieved union, as in transpedicular cement grafting. 2. Materials and methods After obtaining Institutional Review Board approval from the ethics committee of Chang Gung Memorial Hospital, we retrospectively analyzed all consecutive patients with thoracolumbar burst fracture that had undergone surgical intervention at our department from 2009 January to 2012 June. Patients enrolled in the study had to meet the following inclusion criteria: (a) a single-level type-A3 burst fracture according to the AO (Arbeitsgemeinschaft für Osteosynthesefragen) Magerl classification [14]; (b) a level between T11 and L3; (c) having undergone short-segment posterior instrumentation without posterior fusion or posterolateral fusion; (d) the fractured vertebra was augmented by 2 additional screws or injectable calcium sulfate/phosphate cement; (e) having received at least 2 years of follow-up with radiographic and clinical data (Table 1). The American Spinal Injury Association (ASIA) impairment scale was used to evaluate patients’ preoperative and final neurologic status. The final clinical results were assessed using the Denis scale [15], which is a 5-point scale that includes both work and pain scales. Table 2 described details of pain and work scale. Lower points on Denis scale represented a better clinical outcome. Preoperative computed tomography (CT) of the spine was used to measure the degree of canal encroachment; the percentage of

Pain Scale P1: no pain P2: occasional pain not requiring medication P3: moderate pain requiring occasional medication P4: moderate to severe pain requiring frequent medication P5: constant incapacitating pain requiring chronic medication Work Scale W1: returned to previous employment W2: capable but did not return to previous employment W3: unable to return to previous employment and currently employed in a different full-time job W4: unable to return to previous employment and currently working part-time or frequently absent from work because of pain W5: completely disabled and unable to work Source: Denis F et al. Clin Orthop Relat Res. 1984;189:142-9 [15]. P = pain, W = work.

canal encroachment was calculated using the formula developed by Mumford et al. [16]. Preoperative, immediately postoperative, and final follow-up plain radiographs were analyzed. Sagittal kyphosis was measured from the superior endplate of the cephalic intact vertebra to the inferior endplate of the caudal intact vertebra. The normal height of the fractured vertebrae on lateral radiographs was determined by averaging the heights of the adjacent cephalic and caudal vertebrae. The percentage of anterior height of the fractured vertebra was calculated as the anterior height of the injured vertebra/the estimated normal anterior height of the injured vertebra × 100%. The severity of the fractured level was scored according to the load-sharing classification (LSC) using preoperative X-rays and CT scans [17]. The LSC determines the fractured body according to three components: (1) community of the body (2) apposition of the fractured fragments (3) deformity correction after surgery. Each component is classified from one point to three points. The total LSC scores are ranged from three points to nine points; more points represent more severity of the fractured vertebrae (Fig. 1). The surgery was considered a failure if the implant failed or if radiographs obtained at the final follow-up showed an increase of 10◦ or more in sagittal kyphosis compared to the local kyphosis angle measured immediately after surgery. Demographic data, including age, sex, injury level, estimated blood loss, operation time, duration of admission, time between injury and surgery, fluoroscope time, and associated injuries, were collected. 2.1. Surgical procedure All patients were under general anesthesia and placed in the prone position on the four-poster. General anesthesia was performed basically with administration of Propofol, Fentanyl, and Cisatracurium at our institute. Continuous arterial pressure monitoring and somatosensory evoked potential were used to monitor every patient. Posture and manual reduction were applied first. After the skin was draped and prepared, a standard posterior midline approach was used to explore the spine (from one-above to one-below the fractured vertebra). 2.2. Allpedicle screw method (group 1) Monoaxial pedicle screws were inserted in the above and the fractured vertebra, polyaxial pedicle screws were in the below vertebra. The positions of the screws were confirmed by the C-arm; screws in the fractured vertebra had to be parallel to the healthy end-plate. The rod was bent slightly at a lordotic angle and was connected to the screws; initial fracture reduction could be obtained by a cantilever maneuver. Then a distraction force was applied using

58

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63

Fig. 1. Illustration of the load sharing classification (LSC) for the thoracolumbar fractures. The LSC scores are determined by three components: comminution of the fractured body; apposition of fragments; kyphosis correction after surgery. Comminution of the body is assigned 1 point (<30% on sagittal computed tomography), 2 points (30%–60%), 3 points (>60%). Apposition of fragments is assigned 1 point: (minimal displacement on axial computed tomography), 2 points (2 mm displacement involving <50% of the cross section of the body), 3 points (2 mm displacement involving >50% of the cross section of the body). Kyphosis correction is assigned 1 point (kyphotic correction 3◦ ), 2 points (kyphotic correction 4◦ –9◦ ), 3 points (kyphotic correction 10◦ )

the spreader forceps to further correct local kyphosis and restore the anterior body height. 2.3. Fracture augmentation by injectable calcium sulfate/phosphate cement method (group 2) Pedicle screws were inserted into the vertebra 1 level above and 1 level below the fractured vertebra. After connecting the rods and screws, a distraction force was applied using spreader forceps to restore lordosis and body height. A trocar in a cannula was inserted into the defect of the fractured body through the pedicle. An assistant prepared the injectable calcium sulfate/phosphate cement (PRO-DENSE, Wright Medical Technology, Arlington, Tennessee, USA) when the cannula reached the optimal position. After moving the trocar, the cement (about 5 ml) was injected into the fractured vertebra through the cannula under continuous fluoroscope monitoring.The rods were connected with a cross-link. On the second postoperative day, the patients were encouraged to sit and began rehabilitation in preparation of ambulation. All patients were protected with a Taylor brace for 3 months. 2.4. Statistics The paired t-test was used to analyze differences between preoperative, postoperative, and final follow-up radiographic data within each group. The Mann-Whitney test was used to analyze numerical data between the 2 groups. Fisher’s exact test was used for categorical variables. The level of statistical significance was set at p < 0.05. 3. Results There were sixty-eight patients with one-level thoracolumbar burst fracture underwent short-segment instrumentation by the authors during this study period. Ten patients were treated with

additional posterior fusion by iliac autogenous bone graft, two patients were loss of follow-up, and totally twelve cases were excluded. Finally, 56 patients met the inclusion criteria and were enrolled into the study. 38 were men and 18 were women, and the average age at surgery was 43.4 ± 11.1 years. The mean operation time of these 56 patients was 166.4 ± 43.11 min. Twenty-seven patients underwent a six-screw construct (group 1) (Fig. 2), and 29 were treated with four-screw short-segment instrumentation and injectable calcium sulfate/phosphate cement to reinforce the fracture vertebra (group 2) (Fig. 3). There were no statistically significant differences in gender, age, injury level, injury mechanism, and associated injuries between the 2 groups. Preoperative neurologic status, distribution was also similar in both groups. Patients in group 2 had a shorter waiting period for surgery (4.4 ± 2.6 vs. 2.5 ± 1.4 days, p = 0.005), and total hospital stay was decreased (11.9 ± 9.1 vs. 9.1 ± 3.4 days, p=0.0.002). Estimated blood loss and operation time were less in group 1 (126.2 ± 89.7 vs. 267.6 ± 126.1 ml, p < 0.001; 141.2 ± 48 vs. 189.8 ± 16.4 min, p < 0.001) (Table 3). In group 1, the average preoperative spinal canal encroachment as determined by CT was 48.0% ± 16.1%. The mean preoperative kyphotic angle was 20.9◦ ± 6.2◦ , which was corrected to 6.0◦ ± 4.1◦ immediately after surgery. This was a correction of 14.9◦ ± 4.9◦ (p < 0.001). At the final follow-up, the local sagittal angle became 9.9◦ ± 4.6◦ , and loss of kyphosis correction was 4.2◦ ± 3.5◦ . There was still a statistically significant 11◦ correction from injury to the final visit (p < 0.001). The mean preoperative anterior body height was 49.4% ± 11.7%, which improved to 84.9% ± 10.0% immediately after surgery (p < 0.001). The anterior body height was restored 35.5% ± 13.3% with surgery. At the final follow-up, the anterior body height had collapsed significantly to 76.0% ± 12.0%. Compared to the preoperative status, there was still a statistically significant 26.6% mean restoration at the final visit (p < 0.001). The average fluoroscope time was 26.7 s. The mean LSC score was 6.3.

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63

59

Fig. 2. A 42-year-old male with L1 burst fracture underwent a six-screw construct. A. Preoperative radiograph showed 27.2◦ local kyphosis and 49% anterior height collapse. B. Immediate postoperative radiograph revealed 8.4◦ local kyphosis and 14% anterior body height collapse. C. Two years after surgery, the radiograph showed local kyphosis had deteriorated to 11.3◦ and anterior body height collapse increased to 17%.

Fig. 3. A 46-year-old male with an L2 burst fracture underwent a four-screw construct and fractured vertebra augmentation with injectable calcium sulfate/phosphate cement. A. Preoperative radiograph showed 24.8◦ local kyphosis and 41% anterior height collapse. B. Immediate postoperative radiograph revealed 2.6◦ local kyphosis and 2% anterior body height collapse. C. Two years after surgery, the radiograph showed L2 body union. Local kyphosis deteriorated to 7.3◦ and anterior body height collapse increased to 11%.

In group 2, preoperative CT demonstrated that the mean spinal canal encroachment was 45.8% ± 21.2%. The average preoperative local kyphosis angle was 18.6◦ ± 6.0◦ , which was corrected to 3.7◦ ± 5.3◦ immediately after surgery. The kyphosis correction was 14.8◦ ± 6.4◦ by operation (p < 0.001). The final mean local kyphosis was 10.0◦ ± 8.2◦ . Loss of kyphosis correction was 6.3◦ ± 5.7◦ . However, there was still a statistically significant 8.5◦ of correction between injury and the final visit (p < 0.001). The average preoperative anterior body height was 54.3% ± 13.6%, which improved to 94.9% ± 7.6% immediately after surgery. Postoperative anterior body height was restored by 40.6% ± 15.2%. The final anterior body was 81.6% ± 14.0%, and the average loss of body correction was 13.3% ± 16.5%. There was still a 27.3% anterior body height acquisition between injury and final follow-up (p < 0.001). The mean LSC score in group 2 was 6.4. The average fluoroscope time was 53.0 s. These 2 groups had similar LSC scores and showed no significant differences in radiographic data except postoperative local kyphosis (6.0◦ ± 4.1◦ vs. 3.7◦ ± 5.3◦ , p = 0.047) and postoperative anterior body height (84.9% ± 10.0% vs. 94.9% ± 7.6%, p < 0.001). The fluoroscope time of the group 2 was also significantly longer than that of the group 1 (53.0 ± 9.2 vs. 26.7 ± 4.3 s, p < 0.001), because con-

tinue fluoroscope monitor was necessary for performing fractured body augmentation with injectable artificial bone substitute. The radiographic data comparison between the two groups was demonstrated in Table 4. In group 1, the failure rate was 11.1% (3/27), including 1 patient with implant failure (screw breakage) and2with loss of correction >10◦ . These 3 patients had an LSC score of 7. In group 2, the failure rate was 27.6% (8/29): all 8 patients had broken rods (6 at the upper part of the rod, 2 at the lower part of the rod), and 5 of these 8 patients sustained a loss of kyphosis correction >10◦ . Figs. 4 and 5 demonstrate a typical case with implant failure in group 1 and group 2, respectively. Based on LSC scores, 1 was 6, 4 were 7, 2 were 8, and 1 was 9. The operation records showed that 5 of these 8 patients had a posterior complex injury. For both groups, the failure rate was 7.1% (2/28) when the LSC score was less than or equal to 6 and increased to32.1% (9/28) when the LSC score was equal to or over 7 (p = 0.026). 22 patients in group 1 (22/27, 81%) and 25 patients in group 2 (25/29, 86%) had undergone a second surgery to remove their spine implants. In group 1, the mean pain score was 1.5 ± 0.8, and the mean work score was 1.7 ± 0.9 finally. In group 2, the average

60

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63

Table 3 Patient Demographic Data. Characteristic

Group 1 (N = 27)

Group 2 (N = 29)

P values

Age (years)

41.1 ± 11.2

45.5 ± 10.7

Gender Female Male

11 16

7 22

0.133 0.184

Level T11 T12 L1 L2 L3

0 6 15 4 2

0 2 14 11 3

0.155

Hospital stay (days) Injury to operation interval (days) Operation time (min) Blood loss (c.c.)

11.9 ± 9.1 4.4 ± 2.6 141.2 ± 48.7 126.2 ± 89.7

9.1 ± 3.4 2.5 ± 1.4 189.8 ± 16.4 267.6 ± 126.1

0.005 0.002 <0.001 <0.001

Mechanism Fall MVA Struck

18 8 1

19 9 1

0.993

Associated injury Yes No

12 15

11 18

0.621

MVA = motor vehicle accident. Table 4 Radiographic Data of Surgery. Parameter

Group 1 (N = 27)

Group 2 (N = 29)

P values

Failure rate Preoperative canal encroachment (%)

3/27 (11.1%) 48.0 ± 16.1

8/29 (27.6%) 45.8 ± 21.2

0.121 0.884

Local kyphosis (degree) Preoperative Postoperative Final Correction by surgery Loss of correction at final Preoperative vs. Postoperative Postoperative vs. Final Preoperative vs. Final

20.9 ± 6.2 6.0 ± 4.1 9.9 ± 4.6 14.9 ± 4.9 4.2 ± 3.5 P < 0.001 P < 0.001 P < 0.001

18.6 ± 6.0 3.7 ± 5.3 10.0 ± 8.2 14.8 ± 6.4 6.3 ± 5.7 P < 0.001 P < 0.001 P < 0.001

0.125 0.047 0.555 0.705 0.232

Anterior body height (%) Preoperative Postoperative Final Correction by surgery Loss of correction at final Preoperaitve vs. Postopersitve Postoperaitve vs. Final Preoperaitve vs Final Fluoroscope time (seconds)

49.4 ± 11.7 84.9 ± 10.0 76.0 ± 12.0 35.5 ± 13.3 8.9 ± 6.4 p < 0.001 p < 0.001 p < 0.001 26.7 ± 4.3

54.3 ± 13.6 94.9 ± 7.6 81.6 ± 14.0 40.6 ± 15.2 13.3 ± 16.5 p < 0.001 P < 0.001 P < 0.001 53.0 ± 9.2

0.353 <0.001 0.027 0.076 0.464

Load sharing score 3 4 5 6 7 8 9

0 2 2 12 11 1 0

0 2 4 7 11 3 1

pain scale and work scale were 1.4 ± 0.6 and 1.6 ± 1.0, respectively. No statistically significant difference in pain and work scores (p = 0.706 and 0.854, respectively) was observed between the 2 groups. (Table 5) According to the ASIA grading system, 1, 3, and 23 patients in group 1 were classified as grade C, D, and E; 2, 4, and 23 patients in group 2 were classified as grade C, D, and E before surgery. No one sustained a neurologic deterioration due to surgery (Table 6).

<0.001

0.759

Two patients had wound infections; one was in group 1 and was being treated successfully with antibiotics, and the other was in group 2 and was treated with surgical debridement 3 weeks after initial surgery. Although the wound healed finally, hardware failure was noted at the 6-month follow-up and the final radiographic data was worse than the preoperative data.

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63

61

Fig. 4. A case with implant failure in group 1; a 43-year-old male with a T12 burst fracture underwent a six-screw construct. A. Preoperative radiograph showed 26.3◦ local kyphosis and 52% anterior height collapse. B. Immediate postoperative radiograph revealed 8.1◦ local kyphosis and 14% anterior body height collapse. C. The final radiograph showed the L1 screw was broken. Local kyphosis deteriorated to 16.3◦ and anterior body height collapse increased to 39%.

Fig. 5. A patient with implant failure in group 2; a 38-year-old female with an L1 burst fracture underwent a four-screw construct and fractured vertebra augmentation with injectable calcium sulfate/phosphate cement. A. Preoperative radiograph showed 23.2◦ local kyphosis and 61% anterior height collapse. B. Immediate postoperative radiograph revealed 4.7◦ local kyphosis and 3% anterior body height collapse. C. The final radiograph showed an L1 body vacuum phenomenon with the T12-L1 rod broken.

Table 5 Clinical Outcomes Using Denis Scale.

Table 6 Distribution of neurologic status according ASIA impairment scale.

Group 1 (N = 27)

Group 2 (N = 29)

P values

Pain scale 1 2 3 4 5

17 7 2 1 0

17 11 1 0 0

0.706

Work scale 1 2 3 4 5

14 10 1 2 0

17 9 1 1 1

0.854

Parameter

A

B

C

D

E

Group 1 (N = 27) Preoperative Final

0 0

0 0

1 0

3 1

23 26

Group 2 (N = 29) Preoperative Final

0 0

0 0

2 1

4 1

23 27

ASIA = American Spinal Injury Association.

4. Discussion Treatments for thoracolumbar burst fractures can be nonsurgical or surgical method. The absolute indications for surgical

treatment of a thoracolumbar burst fracture include progressive neurologic deficit, conus medullaris or cauda equine syndrome due to bony impingement, or severe spinal instability. For a patient who is neurologically intact and has an intact posterior ligament complex, nonsurgical treatment with bed rest and body orthosis can advocate initially. However, there are still many patients who are initially undergoing nonsurgical treatment fail and then shift to surgical treatments. Hitchon et al. demonstrated a study that sixty-eight neurologically intact patients with thoracolumbar burst

62

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63

fractures were treated with nonoperative treatment initially, but 18 patients failed nonoperative management then shifted to instrumentation [18]. The authors found those who failed nonsurgical management were significantly more kyphotic and stenotic residual canal than those successfully treated nonoperatively (8◦ versus 3◦ ; 52% versus 63%, respectively). In the current study, the average preoperative local kyphosis was 20.9◦ in the group 1 and 18.6◦ in the group 2; the mean percentage of residual canal was 52% in the group 1 and 54% in group 2. We believed our cases in this study should be treated surgically to prevent failure of nonsurgical treatment. Furthermore, the authors’ country is located in a tropical area with hot climate; many patients can’t tolerate the nonoperative treatment of body orthosis for thoracolumbar fracture. Controversy still exists regarding the use of fusion or non-fusion with posterior short-segment instrumentation for thoracolumbar burst fractures. The traditional reasons for additional posterior fusion include decreasing implant failure incidence and longterm maintenance of alignment [19,20]. However, a few studies have demonstrated that non-fusion with posterior short-segment instrumentation obtained radiographic and clinical results similar to those of posterior short-segment instrumentation with fusion for thoracolumbar burst fractures [21,22]. The additional advantages of the non-fusion method include prevention of donor-site morbidity and a decrease in operating time and blood loss. Dai et al. examined 36 patients with Denis type-B thoracolumbar burst fracture with an average LSC score of 4.2 that underwent short-segment instrumentation (four-screw construct) without fusion, and no implant failure was noted in the patients [21]. In our study, the mean LSC score was 6.3 in group 1 and 6.4 in group 2, which means our patients’ severity was higher than that of Dai et al; all patients with an LSC score of 8 or 9 developed implant failures but the failure rate was dropped when the LSC score was 7 or less. We believe that four-screw fixation without fusion is only for those thoracolumbar bursts with less severity (LSC score 3, 4, 5). By additional augmentation with two screws or transpedicular grafting inside the fractured body (our method), it was still safe for the severity of LSC score 6 or 7. Without anterior support, posterior instrumentation might fail due to a large anterior defect created during the application of distraction, although modern pedicle screws have high pull-out and cut-out strength and can resist high levels of stress. Transpedicular augmentation of the fractured vertebra seems to be an ideal method to maintain alignment and vertebra height, and prevent implant failure. Marco et al. claimed that posterior short-segment instrumentation plus calcium phosphate cement for patients with burst fractures could achieve satisfactory clinical results and a low instrumentation failure rate (5.3%) [23]. Shen et al. and Liao et al. successfully used posterior short-segment instrumentation plus injectable calcium sulfate cement for thoracolumbar burst fractures with an implant failure rate from 0% to 5% [10,24]. Our study showed the implant failure rate in group 2 (using calcium sulfate/phosphate cement) was higher than previous studies (27.6%). The bone substitute cement (PRO-DENSE) we used in the present study had been reported to be used in surgically managing patients with primary bone tumors and femoral head necrosis [25,26]. PRO-DENSE contains a matrix of CaSO4/CaPO4 embedded with BTCP granules [27]. This compound theoretically provides better biomechanical stability and a better osteoconductive environment for new bone formation than a pure calcium sulfate or calcium phosphate cement. Why were our results worse than those of previous studies using calcium sulfate cement or calcium phosphate cement? The average anterior height correction was 41% of all patients in group 2, and 45% for the 8 patients with implant failure in group 2. Liao’s results showed a mean of 36% anterior body height correction in the calcium sulfate cement group [10]. Since the bone defect created inside the injured vertebra should be refilled with adequate bone substitute grafting to prevent early collapse and

stimulate bone growth, we deduced that the size of the defect of the fractured vertebra in group 2 in our study was too large, the amount of calcium sulfate/phosphate cement was too little to maintain body height, and the posterior instrumentation was inherently too much loading, which resulted in implant failure. The insertion of additional screws at the fractured vertebra was proposed as an alternative for augmentation of the fracture level. Guven et al. reported 18 of 54 patients with thoracolumbar burst fracture with an average LSC score of 6 that underwent a six-screw construct; the clinical and radiographic results were comparable to those of patients undergoing long instrumentation [28]. Farrokhi et al. showed that 6% (2/38) of patients with thoracolumbar burst treated with a six-screw construct had an implant failure [29]. Pellise et al. examined 72 patients that underwent a six-screw construct for thoracolumbar burst fracture and noted no implant failure, but loss of correction was dramatically increased when the LSC score was 7 [30]. The common point of these 3 studies was that posterior fusion was performed with autogenous bone graft, so blood loss was up to 400 ml. In contrast, the mean blood loss was 126 ml in group 1 of the current study. Wang et al. demonstrated a study that they used one additional pedicle screw inserted into one side of the fractured vertebra; the authors claimed that only 3 patients (8.3%) in the non-fusion group had screw breakage [13]. In our study, the average loss of correction in group 1 (six-screw construct) was 4.2◦ , but only one patient had an implant failure. In addition, two other patients had loss of correction >10◦ at the final follow-up. The LSC score of these three patients was 7, so some authors suggested that short-segment pedicle screws with 2 additional augmenting screws in the injured vertebra without fusion should be reserved for those with thoracolumbar burst fracture with an LSC score 6 [31]. Furthermore, in those patients with osteoporotic vertebrae or with underlying neuromuscular disease, six-screw short-segment fixation was not suggested to be used because the screws could not be held well. Under these conditions, long instrumentation (at least eight-screw construct) would be an ideal choice to prevent implant failure or recurrent kyphosis. All patients in the current study had a Magerl type A3 thoracolumbar burst fracture. Posterior complex, including posterior ligament and facet joint is rarely disrupted in this type A3 fracture, which is why we did not routinely perform posterior or posterolateral fusion for these patients. However, five patients in these eight patients in both groups with an implant broken and/or with a loss of correction >10◦ (the definition of failure surgery) had a posterior complex injury, which means integrity of the posterior complex plays a role to prevent implant failure and maintain alignment. Therefore, we suggest the following: if posterior element (facet joint or interspinal ligament) injuries are apparent during surgery, posterior fusion should be added. Whether posterior implant removal after union of fractured vertebrae is beneficial or not remains uncertain. The advantages of implant removal after fracture fusion include decrease of implantrelated pain and increase of non-fused motion segment. In contrast, the disadvantages of implant removal include loss of correction and surgical related wound problems. Reviewing literatures in recent 5 years, most authors advocated for removing implant electively because of lower presence and level of pain, improvement of function [32,33]. In Taiwan culture, furthermore, many people believe it is not good for one’s health with metal inside his/her body. That was why so many patients in this study (47/56, 84%) accepted a second surgery to remove their implants.

5. Conclusions The six-screw construct was easier to perform with less blood loss compared to the four-screw construct with fractured verte-

J.-C. Liao, K.-F. Fan / Clinical Neurology and Neurosurgery 153 (2017) 56–63

bra augmentation by injectable calcium sulfate/phosphate cement. When the LSC score 6, both methods can be used safely. With an LSC 7, both groups would have a higher incidence of implant failure or re-kyphosis, especially the four-construct group with body augmentation by injectable calcium sulfate/phosphate cement. Conflicts of interest I (Dr Jen-Chung Liao) and Dr. Kuo-Fon Fan declared that no funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript. Funding No funding was received for this study. Ethical approval This study was performed after obtaining approval from the institutional review board of Chang Gung Memorial Hospital (No. 102-2142B). Informed consent We gave all the patients a written consent to the use of data for research. Acknowledgement I would like to thank Julie Huang for editing this article. References [1] D.R. Benson, J.K. Burkus, P.X. Montesano, T.B. Sutherland, R.F. McLain, Unstable thoracolumbar and lumbar burst fractures treated with the AO fixateur interne, J. Spinal Disord. 5 (1992) 335–343. [2] N.R. [2[.Crawford, C.A. Dickman, Construction of local vertebral coordinate systems using a digitizing probe, Tech. Note: Spine 22 (1997) 559–563. [3] P.W. Hitchon, W. He, S. Viljoen, N.S. Dahdaleh, R. Kumar, J. Noeller, J. Torner, Predictors of outcome in the non-operative management of thoracolumbar and lumbar burst fractures, Br. J. Neurosurg. 28 (2014) 653–657. ˜ [4] M. Cahueque, A. Cobar, C. Zuniga, G. Caldera, Management of burst fractures in the thoracolumbar spine, J. Orthop. 13 (2016) 278–281. [5] H. Aono, H. Tobimatsu, K. Ariga, M. Kuroda, Y. Nagamoto, S. Takenaka, M. Furuya, M. Iwasaki, Surgical outcomes of temporary short-segment instrumentation without augmentation for thoracolumbar burst fractures, Injury 47 (2016) 1337–1344. [6] C. Li, J. Pan, Y. Gu, J. Dong, Minimally invasive pedicle screw fixation combined with percutaneous vertebroplasty for the treatment of thoracolumbar burst fracture, Int. J. Surg. 36 (Pt. A) (2016) 255–260. [7] A.D. Steffee, R.S. Biscup, D.J. Sitkowski, Segmental spine plates with pedicle screw fixation: a new internal fixation device for disorders of the lumbar and thoracolumbar spine, Clin. Orthop. Relat. Res. 203 (1986) 45–53. [8] U. Müller, U. Berlemann, J. Sledge, O. Schwarzenbach, Treatment of thoracolumbar burst fractures without neurologic deficit by indirect reduction and posterior instrumentation: bisegmental stabilization with monosegmental fusion, Eur. Spine J. 8 (1999) 284–289. [9] J.A. Goulet, L.E. Senunas, G.L. DeSilva, M.L. Greenfield, Autogenous iliac crest bone graft: complications and functional assessment, Clin. Orthop. Relat. Res. 339 (1997) 76–81. [10] J.C. Liao, K.F. Fan, G. Keorochana, W.J. Chen, L.H. Chen, Transpedicular grafting after short-segment pedicle instrumentation for thoracolumbar burst fracture: calcium sulfate cement versus autogenous iliac bone graft, Spine 35 (2010) 1482–1488.

63

[11] P. Korovessis, A. Hadjipavlou, T. Repantis, Minimal invasive short posterior instrumentation plus balloon kyphoplasty with calcium phosphate for burst and severe compression lumbar fractures, Spine 33 (2008) 658–667. [12] R.P. Norton, E.L. Milne, D.N. Kaimrajh, F.J. Eismont, L.L. Latta, S.K. Williams, Biomechanical analysis of four- versus six-screw constructs for short-segment pedicle screw and rod instrumentation of unstable thoracolumbar fractures, Spine J. 14 (2014) 1734–1739. [13] S.T. Wang, H.L. Ma, C.L. Liu, W.K. Yu, M.C. Chang, T.H. Chen, Is fusion necessary for surgically treated burst fractures of the thoracolumbar and lumbar spine? A prospective, randomized study, Spine 31 (2006) 2646–2652. [14] F. Magerl, M. Aebi, S.D. Gertzbein, J. Harms, S. Nazarian, A comprehensive classification of thoracic and lumbar injuries, Eur. Spine J. 3 (1994) 184–201. [15] F. Denis, G.W. Armstrong, K. Searls, L. Matta, Acute thoracolumbar burst fractures in the absence of neurologic deficit: a comparison between operative and nonoperative treatment, Clin. Orthop. Relat. Res. 189 (1984) 142–149. [16] J. Mumford, J.N. Weinstein, K.F. Spratt, V.K. Goel, Thoracolumbar burst fractures. The clinical efficacy and outcome of nonoperative management, Spine 18 (1993) 955–970. [17] T. McCormack, E. Karaikovic, R.W. Gaines, The load sharing classification of spine fractures, Spine 19 (1994) 1741–1744. [18] P.W. Hitchon, K. Abode-Iyamah, N.S. Dahdaleh, C. Shaffrey, J. Noeller, W. He, T. Moritani, Nonoperative management in neurologically intact thoracolumbar burst fractures: clinical and radiographic outcomes, Spine 41 (2016) 483–489. [19] J.W. Parker, J.R. Lane, E.E. Karaikovic, R.W. Gaines, Successful short-segment instrumentation and fusion for thoracolumbar spine fractures: a consecutive 41/2-year series, Spine 25 (2000) 1157–1170. [20] B.P. Qian, Y. Qiu, B. Wang, Y. Yu, Z.Z. Zhu, Effect of posterolateral fusion on thoracolumbar burst fractures, Chin. J. Traumatol. 9 (2006) 349–355. [21] L.Y. Dai, L.S. Jiang, S.D. Jiang, Posterior short-segment fixation with or without fusion for thoracolumbar burst fractures: a five to seven-year prospective randomized study, J. Bone Joint Surg. Am. 91 (2009) 1033–1041. [22] N. Jindal, S.S. Sankhala, V. Bachhal, The role of fusion in the management of burst fractures of the thoracolumbar spine treated by short segment pedicle screw fixation: a prospective randomised trial, J. Bone Joint Surg. Br. 94 (2012) 1101–1106. [23] R.A. Marco, B.C. Meyer, V.P. Kushwaha, Thoracolumbar burst fractures treated with posterior decompression and pedicle screw instrumentation supplemented with balloon-assisted vertebroplasty and calcium phosphate reconstruction, J. Bone Joint Surg. Am. 92 (Suppl. 1 Pt. 1) (2010) 67–76, Surgical technique. [24] Y.X. Shen, P. Zhang, J.G. Zhao, W. Xu, Z.H. Fan, Z.F. Lu, L.B. Li, Pedicle screw instrumentation plus augmentation vertebroplasty using calcium sulfate for thoracolumbar burst fractures without neurologic deficits, Orthop. Surg. 3 (2011) 1–6. [25] N. Evaniew, V. Tan, N. Parasu, E. Jurriaans, K. Finlay, B. Deheshi, M. Ghert, Use of a calcium sulfate-calcium phosphate synthetic bone graft composite in the surgical management of primary bone tumors, Orthopedics 36 (2013) e216–222. [26] R. Civinini, P. De Biase, C. Carulli, F. Matassi, L. Nistri, R. Capanna, M. Innocenti, The use of an injectable calcium sulphate/calcium phosphate bioceramic in the treatment of osteonecrosis of the femoral head, Int. Orthop. 36 (2012) 1583–1588. [27] R.M. Urban, T.M. Turner, D.J. Hall, N. Inoue, S. Gitelis, Increased bone formation using calcium sulfate-calcium phosphate composite graft, Clin. Orthop. Relat. Res. 459 (2007) 110–117. [28] O. Guven, B. Kocaoglu, M. Bezer, N. Aydin, U. Nalbantoglu, The use of screw at the fracture level in the treatment of thoracolumbar burst fractures, J. Spinal Disord. Tech. 22 (2009) 417–421. [29] M.R. Farrokhi, A. Razmkon, Z. Maghami, Z. Nikoo, Inclusion of the fracture level in short segment fixation of thoracolumbar fractures, Eur. Spine J. 19 (2010) 1651–1656. [30] F. Pellisé, D. Barastegui, A. Hernandez-Fernandez, S. Barrera-Ochoa, J. Bagó, D. Issa-Benítez, E. Cáceres, C. Villanueva, Viability and long-term survival of short-segment posterior fixation in thoracolumbar burst fractures, Spine J. 15 (2015) 1796–1803. [31] S.B. Ko, S.W. Lee, Result of posterior instrumentation without fusion in the management of thoracolumbar and lumbar unstable burst fracture, J. Spinal Disord. Tech. 27 (2014) 189–195. [32] S.I. Stavridis, P. Bücking, S. Schaeren, B. Jeanneret, K.J. Schnake, Implant removal after posterior stabilization of the thoraco-lumbar spine, Arch. Orthop. Trauma Surg. 130 (2010) 119–123. [33] T. Jentzsch, V. Gomes de Lima, B. Seifert, K. Sprengel, C.M. Werner, The benefits of elective spinal implant removal: a retrospective study of 137 patients, Eur. Spine J. 25 (2016) 856–864.