Article Text

Download PDFPDF

Case series
Treating traumatic thoracolumbar spine fractures using minimally invasive percutaneous stabilization plus balloon kyphoplasty: a 102-patient series
  1. Henri Salle1,
  2. Alexandre Meynard1,
  3. Emilie Auditeau2,
  4. Clément Gantois1,
  5. Aymeric Rouchaud3,4,
  6. Charbel Mounayer3,
  7. Patrick Faure1,
  8. François Caire1
  1. 1 Neurosurgery, CHU Limoges, Limoges, France
  2. 2 Epidemiology and Statistical Analysis, CHU Limoges, Limoges, France
  3. 3 Interventional Neuroradiology, CHU Limoges, Limoges, France
  4. 4 University of Limoges, CNRS, XLIM, UMR 7252, Limoges, France
  1. Correspondence to Dr Henri Salle, Neurosurgery, CHU Limoges, Limoges, France; henrisalle1{at}


Background There is no consensus on the treatment for spinal injuries resulting in thoracolumbar fractures without neurological impairment. Many trauma centers are opting for open surgery rather than a neurointerventional approach combining posterior percutaneous short fixation (PPSF) plus balloon kyphoplasty (BK).

Objective To assess the safety and efficacy of PPSF+BK and to estimate the expected improvement by clarifying the factors that influence improvement.

Methods We retrospectively reviewed patients who underwent PPSF+BK for the treatment of single traumatic thoracolumbar fractures from 2007 to 2019. Kyphosis, loss of vertebral body height (VBH), clinical and functional outcomes including visual analog scale and Oswestry disability index were assessed. We examined the overall effects in all patients by constructing a linear statistical model, and then examined whether efficacy was dependent on the characteristics of the patients or the fractures.

Results A total of 102 patients were included. No patient experienced neurological worsening or wound infections. The average rates of change were 74.4% (95% CI 72.6% to 76.1%) for kyphosis and 85.5% (95% CI 84.4% to 86.6%) for VBH (both p<0.0001). The kyphosis treatment was more effective on Magerl A3 and B2 fractures than on those classified as A2.3, as well as for fractures with slight posterior wall protrusion on the spinal canal. A higher postoperative visual analog scale score was predictive of poorer outcome at 1 year.

Conclusions This is the largest series reported to date and confirms and validates this surgical treatment. All patients exhibited improved kyphosis and restoration of VBH. We advise opting for this technique rather than open surgery.

  • balloon
  • spinal cord
  • spine
  • trauma

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Traumatic thoracolumbar spine fractures are commonly seen at spinal trauma centers. In 60% of cases, the fractures areat D12–L1, which is a weak area when the thoracic segment is in kyphosis andthe lumbar segment is in lordosis. There is no consensus on how to treat thesefractures.1 2 Many authors recommended a posterior surgical approach,3 ,4 whereas someuse a subsequent anterior approach to strengthen the anterior column. This treatmentsequence offers good results but is invasive, with major risks and a longhospitalization.5

Inthe 2000s, percutaneous osteosynthesis systems appeared.6 Interest developed incombining this new technique with balloon kyphoplasty (BK) to treat comminutedvertebral fractures with reduction of the anterior column and fixation viaposterior osteosynthesis to obtain sagittal balance. Many studies haveevaluated the effectiveness of this minimally invasive approach,7–9 but theeffects of this treatment on sagittal balance parameters have not been welldocumented due to the small number of patients included in these studies.Moreover, most studies on this topic have included different types offractures, mixing osteoporotic, oncologic, and post-traumatic fractures, whichprobably introduced many biases. In addition, no previous study hasstatistically analyzed the quality of this improved technique, and the variousfactors that explain the difference in improvement.We assessed the effectiveness of this treatment by measuring tworadiological parameters: vertebral body height (VBH) and kyphosis. We alsoestimated improvement by clarifying the factors that influence improvement.


Demographic data

We retrospectively reviewed a series of selected patients who underwent posterior percutaneous short fixation (PPSF) plus BK for the treatment of single traumatic thoracolumbar fractures at our Trauma Center from 2007 to 2019. For each patient, age, sex, and level and type of fracture were recorded. The fracture type was defined according to the Magerl classification.10 The eligibility criteria were neurologically intact patients, with only one vertebral fracture per patient, after spinal injury. The exclusion criteria were neurological impairment (ASIA class A–D) related to trauma, previous thoracolumbar surgery, and multilevel injuries. Osteoporotic fractures, whether or not related to obvious spinal cord injury, and type B1 and B3 ligamentous fractures, as well as C types, which are usually related to neurologic impairment, were also excluded. Moreover, in view of the major instability of this type of fracture, most spine surgeons prefer to add a transverse fixation to the posterior fixation, which makes the percutaneous approach impossible.

Fractures caused by trauma with high kinetics were included unless there was doubt about associated osteoporosis.

Institutional review board approval was obtained for this study (IRB00011687). The protocol for involving human subjects was in accordance with the guidelines of the Declaration of Helsinki, and informed consent was obtained from all participants.

Radiological outcome measures

CT was performed preoperatively and at 1 year postoperatively. Parameters measured included segmental vertebral kyphosis (SVK, the angle formed from the lines drawn on the lower endplate of the adjacent cranial vertebra and the upper endplate of the caudal adjacent intact vertebra) (figure 1) and vertebral body height score (VBH, (the shortest vertebral body height of the fractured vertebra divided by the average posterior body height of the two adjacent intact vertebrae)×100) (figure 1). On axial CT views, posterior wall protrusion (PWP) was measured as the length of the spinal canal of the fractured vertebra divided by the average anteroposterior spinal canal diameter of the two adjacent intact vertebrae (grades I–IV, 0–100%), and CT data were evaluated by a neurosurgeon and orthopedic radiologist who were both blinded. MRI was performed after a CT scan only for type B fractures or if there was concern about an eventual disco-ligamentous lesion.

Figure 1

(A, B) Cannulation of the pedicles of the fractured vertebra: the red dot indicates the lateral edge of the pedicle; the green dot indicates the medial edge of the pedicle; and the dotted yellow line shows the optimal trajectory when crossing the pedicle. (C) Segmental vertebral kyphosis (SVK) is reflected by the angle (in degrees) between lines drawn on the lower endplate of the adjacent cranial vertebra and the upper endplate of the caudally adjacent intact vertebra. Normally, the plates are parallel because non-fractured vertebrae do not exhibit kyphosis. (D) The vertebral body height (VBH) score is given by the shortest body height of the fractured vertebra/the average posterior body height of the two adjacent intact vertebrae×100. Thus, the initial estimated vertebral height is set to 100 and the fractured vertebral height is between 0 and 100. The vertebral body loss is therefore 100−VBH; as a height ratio, this value is dimensionless, similar to the VBH itself. (E) The final result after bent rods were inserted to correct the defect. The perioperative X-ray was taken with the patient in the supine position. (F) Axial CT scan taken 1 year postoperatively; cement fills the entire cavity of the fractured vertebra. (G) Sagittal CT scan taken 1 year postoperatively showing re-expansion of the plates and improvement of the kyphosis.

Clinical outcome measures

Operative data and complications were reported. Surgery duration and the type and volume of cement used were recorded. During hospitalization, clinical evaluations were performed to detect any neurological impairment, sepsis, wound infection, and/or embolic disease. Cement leakage was evaluated radiologically, and hospital duration was measured from admission until just before discharge. We also evaluated the number of times osteosynthesis material was removed.

Functional outcome measures

A visual analog scale (VAS) with scores ranging from 0 to 10 were recorded pre- and postoperatively. The Oswestry disability index11 (ODI) assessed the functional parameters relevant to individual patients, expressed as a percentage score. The ODI ranges from minimum disability (0–20%) to being bedbound (81–100%). The ODI scores were recorded at discharge and 1 year postoperatively.

Surgical procedure

Under general anesthesia, all patients were operated on in the prone position on a radio-transparent operating table by the same neurosurgeon. Bolsters were placed under the thorax and iliac crests to facilitate fracture reduction. Spinal lordosis was accentuated by the position of the operating table. Biplane fluoroscopy control was used intraoperatively in all patients.

The first surgical stage involved kyphoplasty of the fractured vertebra. Balloons were inserted in the middle of the vertebral body under fluoroscopic guidance after cannulation of both pedicles of the fractured vertebra (figure 1). The two balloons were directed toward the middle of the vertebral body, and the positions of the balloons were guided by the type of fracture. Each balloon (right and left) was 20 mL in volume and was progressively inflated until it touched the other one, to obtain good restoration of the fractured endplates. The balloons were inflated during all posterior osteosynthesis procedures to obtain optimal re-expansion of the endplates.

The second surgical stage consisted of percutaneous osteosynthesis. Pedicle screws were inserted under visual control in the adjacent vertebrae after cannulating the pedicles under fluoroscopic guidance (figure 1). All screws were inserted using the Legacy Medtronic system (Medtronic, Minneapolis, Minnesota, USA). Once the screws were in place, inserted bent rods were inserted, with a lordosis correction to obtain the best possible correction. The balloons were progressively deflated at the end of the procedure to obtain the best reduction, and after positioning the scope, polymethyl methacrylate (PMMA) or osteogenic bone cement (calcium phosphate cement) was injected (figure 1).

Statistical analysis

For statistical analysis, we used Microsoft Excel in MS Office. The Student’s t-test was used to compare data derived from kyphosis and VBH score measurements at two points: preoperatively and at 1 year postoperatively. Changes in the two variables (kyphosis and VBH) were evaluated using a linear model. To identify the predictors of these changes we performed multivariate analysis. Differences were considered significant at p<0.05.


Descriptive analysis

The study population included 102 patients (75 men, 27 women) of average age 53 years (range 31–78 years) (figure 2). Vertebrae were fractured at the following locations: T11 (n=4), T12 (n=28), L1 (n=47), L2 (n=15), L3 (n=4), and L4 (n=4). The fractures were classified as types A2–3 (n=4), A3–1 (n=4), A3–2 (n=34), A3–3 (n=37), and B2 (n=23). Tables 1 and 2 summarizes the demographic data.

Figure 2

Left side: Flow chart. Right side: Decisional algorithm developed in our institution based on the post-traumatic fracture morphology, as described by the Magerl classification.

Table 1

Demographic and surgical data with clinical and functional outcomes measures and analysis of kyphosis and VBH at 1 year postoperatively

Table 2

Analysis of kyphosis and VBH: pre- versus post-surgery at 1 year

In 10 of the 102 patients there was a PWP on the spinal canal exceeding 75% of the total diameter of the canal, whereas 21 patients had a 50–75% PWP (tables 1 and 2).

PMMA cement was used in 90 patients and osteogenic cement was used in 12 patients under 35 years of age. There were eight cement leaks with no clinical consequences. No significant difference was shown regarding leakage of the two types of cement (p=0.09). Subsequently, 23 patients underwent osteosynthesis material removal.

The mean duration of hospitalization was 3.71 days (range 2–7 days). The mean volume of cement injected was 13.95 mL (range 6–21 mL). The mean operation duration was 79.48 min (range 53–153 min).

No patient experienced neurological worsening or wound infection and all screws were in good positions.

Analysis of functional scores

Overall, the VAS score significantly improved (p<0.01), decreasing from a preoperative score of 6.7 (range 4–10) to 1.94 (range 1–3). The ODI score decreased from 19.6 (range 5–42) at discharge to 4.3 (range 0–15) at 1 year.

Analysis of kyphosis and VBH

The mean preoperative kyphosis was 17.5° (range 6–30°) versus 4.5° (range 1–12°) at 1 year postoperatively. The difference was significant (p<0.0001). The mean preoperative VBH was 57 (range 34–73) versus 93 (range 82–100) at 1 year postoperatively. This difference was also significant (p<0.0001).

Analysis of the changes in kyphosis and VBH at 1 year postoperatively

All patients improved. Improvement occurred linearly according to fracture severity such that patients with a small degree of kyphosis experienced less amelioration than patients with marked kyphosis (figure 3). The treatment improved the kyphosis and VBH by a mean percentage independent of the initial degree (figure 3).

Figure 3

Left side: Evaluation of the 1-year improvement in kyphosis. Right side: Evaluation of the 1-year improvement in vertebral body height improvement. In both diagrams, the blue dots represent the observed values and the black line is the estimated expected improvement from which an average rate of improvement can be calculated.

The average rates of change were 74.4% (95% CI 72.6% to 76.1%) for kyphosis and 85.5% (95% CI 84.4% to 86.6%) for VBH (p<0.0001). Thus, the treatment was considered effective.

Improvements based on patient and fracture features

Note that the higher the VAS score, the lower the kyphosis improvement (average loss of approximately 2% per VAS point). Concerning kyphosis improvement, the treatment of type A3 and B2 fractures was more effective than that for type A2–3 fractures (estimated difference of 15% for the final rate). Finally, the treatment was more effective when PWP was small (0–25%).

After removing the non-significant effects on height, none of the explanatory variables had an effect on VBH improvement.


To the best of our knowledge, this is the largest published series regarding percutaneous surgical treatment of thoracolumbar post-traumatic fractures using PPSF plus BK. The surgical treatment improved both study variables (VBH and kyphosis) in all patients.

Originally, for the management of compression osteoporotic fractures, percutaneous treatments began with vertebroplasty and kyphoplasty.11 Percutaneous interventional techniques have evolved considerably, with an extended list of indications despite the majority of studies only including osteoporotic and/or oncologic fractures. Many different implant devices are now available.

Interventional minimally invasive percutaneous techniques

Vertebral body stenting (VBS) based on balloon-expandable metallic cages (stents) and cement augmentation allows fracture reduction and vertebral body reconstruction while reducing the risk of cement leakage, but stent-cement complex mobilization has been reported.12

The SpineJack (SJ) system (Stryker Corp, Kalamazoo, Michigan, USA) has shown good results in vertebral augmentation and seems able to correct major structural deformities. Premat et al described significant kyphosis reduction >30% in the majority of cases with significantly improved pain using the SJ device. However, in this small series of 19 patients, secondary adjacent level fractures were noted in 21.1%.13

Recently, a new stent screw-assisted internal fixation (SAIF) technique has been proposed for severe neoplastic and osteoporotic fractures.12 14 This percutaneous technique combines the advantages of VBS cement augmentation and the use of a cannulated pedicle screw to restore the anterior spine and anchor the stents to the posterior spine arch.15 16

All of these percutaneous techniques can be used with additional posterior fixation, but it is sometimes difficult to know if this is really necessary.17 La Barbera et al showed that supplementation of the SAIF technique with posterior fixation only led to a marginal decrease in the strain (5–16%) on the bony structures. However, this was a biomechanical study with no clinical component or patient follow-up.15

Velonakis et al studied a percutaneous technique on Magerl A2 split fractures treated with bone augmentation with polyether ether ketone polymer using the Kiva device (Benvenue Medical, Santa Clara, California, USA). Their series included 21 painful post-traumatic fractures, with only nine osteoporotic patients. This technique was effective for pain and height restoration of the vertebral body.18

Loss of reduction

Fuentes et al described a series of A3 burst fractures treated with BK+PPSF, with a stable correction obtained over time. They described a non-significant loss of correction at final follow-up of 3% of VBH and 2% of kyphosis after 2 years, which seemed to occur during the first months after surgery.19

VBS and SJ have been reported as an alternative to kyphoplasty to reduce the deflation effect and decrease loss of height restoration in compression burst fractures.20 Recent studies seem to show better results in the loss of VBH over time using percutaneous implant devices such as the SJ device versus BK alone.21 22

A biomechanical cadaver study has shown the ability of the SJ device to reduce traumatic burst fracture deformation, and significantly better results with regard to height improvement if the SJ was completed with dorsal instrumentation rather than dorsal instrumentation alone.20 It would be interesting to study SJ+dorsal instrumentation versus SJ alone to examine whether dorsal instrumentation is useful with the new implant device.

It is debatable whether PPSF enables the maintenance of VBH and kyphosis over time when using BK. It would be interesting to compare the loss of VBH and kyphosis over time between implant techniques and BK+PPSF in only post-traumatic fractures, excluding osteoporotic and oncologic fractures. We believe that osteoporotic and oncologic fractures are more likely to lose the restoration achieved over time than proven post-traumatic fractures. Thus, the results of these studies could influence the therapeutic decision-making process for purely post-traumatic fractures.

We thought that the higher the volume of cement injected, the greater the restoration of VBH and SVK could be. However, our study did not show improvement in these two study variables. Perhaps the duration and intensity of balloon inflation is ultimately more decisive for improvement than the volume of cement itself. The quality of the cancellous bone must certainly be taken into account. Indeed, it seems rational to suggest that, if a vertebral body has abundant good-quality cancellous bone, less cement will be needed to restore the VBH and SVK. On the other hand, if the amount of cancellous bone is low, more cement will have to be introduced for vertebral expansion, which can be difficult to obtain.

Posterior wall Protrusion (PWP)

Many authors oppose percutaneous treatment when there is a significant PWP, and many surgeons opt for open treatment to combine a preventive decompression procedure. If percutaneous vertebroplasty is considered a contraindication when PWP is present,23 some authors have thought that BK may seem dangerous in cases of PWP. BK requires balloon inflation for expansion of the vertebra that could promote bone fragment mobilization. Some studies tended to show better control of PWP with percutaneous implant devices.24 25 They even showed a decrease of PWP using SJ or armed kyphoplasty devices. The best explanation would be better distribution of stents in a craniocaudal direction only, which could limit an unintended thrust force against the posterior wall and thus avoid the development of neurologic symptoms.25 Meyblum et al measured a mean preoperative PWP of 6.7 mm versus 6.5 mm postoperatively (p=0.02) using SJ.

However, most of these are osteoporotic fractures and very few are purely post-traumatic. In addition, our study included 31 fractures with PWP >50%. None showed neurological aggravation and we did not observe significant differences in pre- and postoperative PWP.

Secondary adjacent level fracture (SALF)

In our study we did not observe any secondary adjacent level fractures (SALFs) at 1 year, while other studies that have used implant devices have reported significant SALFs as high as 31.4% during follow-up.25 This is higher than the risk of SALF after percutaneous vertebroplasty, estimated at 13.7%.26

This high rate of SALF is possibly due to two factors. First, biomechanically, the use of implants such as an SJ device implies craniocaudal strain on the adjacent vertebral plates causing excessive pressure on the adjacent vertebral bodies. The other aspect is that these studies have included a majority of osteoporotic fractures, certainly with associated bone fragility of the whole spinal axis, that may promote new osteoporotic fractures.

It would be interesting to study these implant devices in a traumatic series only, excluding osteoporotic and oncologic fractures.

Cement leakage

In our study PMMA cement is used, except for young patients who were treated with osteogenic cement. All leaks observed were local and all were asymptomatic (5 anterior or lateral, 2 disc, and 1 canal). In our series, 12 patients <35 years of age received treatment with osteogenic cement. We did not find any significant difference between the two types of cement regarding leakage.

One of the main risks of vertebroplasty is cement leakage, which can lead to serious neurological consequences if the leak is in the vertebral canal or a drainage vein. The use of balloons both reduces the fracture and creates a cavity into which more cement can be injected at much lower pressures than in simple vertebroplasty. Our percentage of leakage (7.84%) was significantly lower than that associated with percutaneous vertebroplasty.27 28

Our cement leakage rate was no higher than with percutaneous implant techniques. Indeed, data with the use of VBS reported a leakage rate of 12.2%,12 while Fuentes et al reported a leakage rate of 11% using the same technique (VBK+PPSF).29

All of these studies used PMMA cement. In our series, osteogenic cement was used in 12 young patients aged <35 years due to its bone-inducing properties and ability to resorb. However, it is more difficult to handle because it is partially radiotransparent and solidifies rapidly.

Which percutaneous treatment should be used?

Our series included four type A3.1 and 34 type A3.2 fractures. Some studies have described good results with kyphoplasty alone for type A3 (burst) fractures.30 31

Currently, there are two main types of treatment for the management of these fractures: open and percutaneous. Open surgery remains the reference treatment in many surgical centers. Many surgical colleagues (orthopedists or neurosurgeons) still treat A3.1 type fractures with open surgery and are not interested in the different percutaneous neurointerventional techniques.

Our study was a retrospective study over 12 years (2007–2019). During these years, percutaneous neurointerventional techniques have considerably evolved.

Our patient care has also evolved with the development of these techniques. Our study shows that BK+PPSF treatment has good and notably functional results for A3.1 and A3.2 fractures, and our own attitude has evolved since 2010 for A3.1 with kyphoplasty treatment alone, and since 2015 for A3.2 with stent implant treatment. This evolution in management explains the low number of A3.1 fractures in our series (only 4 of a total of 102 fractures included). However, no study has demonstrated the superiority of one percutaneous treatment over another.

Our series included 23 patients (22.5%) who had their osteosynthetic material removed. This occurred at approximately 18 months postoperatively in young patients requesting removal. Adapting the indications of percutaneous treatment techniques by valid and robust studies would certainly avoid these ablations of materials.

Even if high-level studies are lacking, we clearly envisage that future management of these fractures will be more percutaneous neurointerventional, and that such new percutaneous techniques will make open surgery obsolete. Finally, we suggest that the real question is not the choice between percutaneous or open, but rather what type of percutaneous treatment to choose for what type of fracture.

Based on our own local experience, we have provided our decisional algorithm for the management of post-traumatic fractures (figure 2).

Limitations of study

This study had some limitations. It was a retrospective single-center study, and the findings would have been more robust if the study had been prospective and multicenter. A control group would have provided a better level of proof. However, the diversity of therapeutic management approaches for this type of fracture, when there are no validated guidelines, would have further complicated comparisons between the new technique and a reference technique, which regardless would have been debatable.

We believe that it would be very instructive for the entire surgical and medical community to develop prospective, comparative, randomized studies of the percutaneous techniques used, based on the type of fracture.


This study is the first to investigate improvements in the loss of VBH and local kyphosis and the factors that influence both. This surgical technique yielded an average rate of improvement of 74.4% for kyphosis and 85.5% for VBH, and patients can now be informed about the expected sagittal parameters.

Although this treatment was effective for all fractures studied, the patients who improved most were those with type A3 and B2 fractures and small PWPs, and who were experiencing severe pain. We believe that this technique has the potential to become a routine alternative for traumatic fractures, particularly A3.3 and osseous B2-type fractures, and thus may replace open surgery.

Ethics statements


The authors thank the patients for contributing to this study, and Nicolas Baradel for statistical analysis.



  • Contributors HS, PF, CM and FC: study conception, design, data acquisition, interpretation, and drafting the manuscript and critically revising it. AM, EA, CG, AR: contributed to data acquisition and critical revisions of the manuscript. All authors approved the final manuscript and are accountable for all aspects of the work.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.