Spinal Canal Decompression for Stage III Kummell’s Disease With “Compressing the Posterior Wall of the Vertebral Body” Guided by Intraoperative Ultrasonography - a Preliminary Clinical Study

Background: There is a lot of controversy regarding the treatment of patients with stage III Kummell’s disease (KD). The purpose of this study is to determine feasibility, ecacy and safety of spinal canal decompression via “compressing the posterior wall of the vertebral body”, guided by real time intraoperative ultrasonography guidance for patients with stage III KD. Methods: We conducted a retrospective study of 11 patients with stage III KD associated with neurological decits who underwent surgical treatment in The Second Hospital, Cheeloo College of Medicine, Shandong University, from September 2017 to January 2019. All patients underwent spinal canal decompression using a special L-shaped compressor (angular compressor) guided by real time ultrasonography to squeeze and compress the posterior wall of vertebral body, supplemented by internal xation and posterior fusion. The primary goal was to determine feasibility, safety and ecacy. Therefore, operating time, Intraoperative blood loss, complications and JOA score, local Cobb angle, ODI score of pre- and post-operative were evaluated. Results: 11 patients (4 males and 7 females) with an average age of 70.2±3.2 years (63-76 years) were included in the study. Mean follow-up period was 17.4 months (12-24 months). Mean operating time was 151±18 min and mean intraoperative blood loss was 202±40ml. Before and after surgery, the local Cobb angle (34.3 °±5.4 °, 11.0 °±4.6 °, respectively), JOA score (12.8±2.1, 22.1±1.9, respectively) and ODI scores (71.4±6.4, 25.5±5.2, respectively) signicantly improved (P <0.05). As a complication, one patient developed CSF leakage after surgery, while in another case transient MEP signal change occurred intraoperatively without neurological decit after surgery. Conclusion: Spinal canal decompression by tamping the retropulsed vertebral body anteriorly, guided by intraoperative ultrasonography can effectively relieve neural compression, signicantly shorten operation time, reduce intraoperative blood loss, and have fewer complications for patients

Introduction KD was rst described by German surgeon Hermann Kummell in the 1890s. It represents a speci c type of osteoporotic compression fracture with or without minor trauma, which occurs most frequently in the thoracolumbar vertebrae. 1 Patients mainly complain about local back pain and tenderness at the injury site. As the disease progresses, kyphosis of the involved segments, potentially followed by neural compression may occur 2 .
Li et al 3 divided KD into three stages according to clinical characteristics: stage I, the loss of vertebral body height is less than 20%, with no adjacent disc degeneration, and patients mainly complain lower back pain; stage II, the loss of vertebral body height is more than 20%, usually accompanied by adjacent disc degeneration and/or instability of the fractured vertebral body. These patients primarily present with radiculopathy; stage III, the collapsed posterior wall of vertebral body is displaced dorsally and protrudes into the spinal canal. The patients typically present with lower back pain, often accompanied by severe neurological de cits.
For patients with stage I and II KD, PKP or PVP procedure are usually recommended [4][5][6][7] . However, there is a lot of controversy regarding the treatment of patients with stage III KD. In recent years, many authors have reported different treatment options.
These surgical methods mainly differ in their approach, including anterior or posterior decompression alone, or a combination. Additionally, anterior and middle column xation as well as spinal canal decompression can be achieved via the anterior approach. However, in patients with osteoporosis, the stability of anterior column xation is weaker than that of posterior pedicle screw xation, and anterior surgery is more invasive and with a higher risk of internal organ inujury 8 . Hence, decompression and fusion via a posterior approach has become the treatment of choice for patients suffering from stage III KD 9 . Historically, surgery via a posterior approach also has its shortcomings: due to the fractures being old and the posterior wall of the vertebral body displacing into the spinal canal, satisfactory decompression of the spinal canal is more di cult. Some scholars have adopted bilateral pedicle entry to remove the displaced vertebrae, followed by the application of bone graft to support the anterior and middle columns. This method can obtain complete spinal decompression 2 . However, the vertebral body and posterior structure of the lesion vertebrae are seriously destroyed in the process, compromising the structural stability of the spine. In addition, slow fusion of the bone due to the nature of osteoporosis often results in internal xation failure. Meanwhile, this strategy is more invasive for the patient with more blood loss, longer surgery time, and more complications, making the procedure very controversial 2 .
By considering the advantages and disadvantages of current surgical methods for stage III KD, our study combined intraoperative real time ultrasonography guidance with neurophysiology monitoring, avoiding neural elements irritation during anterior displacement of the posterior wall to the ventral side of the spinal canal with a special L-shaped compressor (Trauson, China). Additionally, we also corrected obvious residual kyphosis with a Smith-Peterson osteotomy (SPO) or Ponte osteotomy to achieve better decompression and corrective results in selected patients. Ultrasound provides protection of the neural elements and immediate feedback of successful decompression.

Materials And Methods
This study was approved by the Ethics Committee of our hospital and conducted according to the principles of the Declaration of Helsinki (Ethical principles for Medical Researches Involving Human Subjects).
We retrospectively collected data on patients with neurological de cits suffering from stage III KD who underwent surgical treatment at our hospital from September 2017 to January 2019.
Inclusion criteria were as follows: stage III KD diagnosed by imaging; neural compression due to the retropulsed posterior wall of the collapsed vertebral body rather than degenerative disc; bone mineral density examination suggesting severe osteoporosis (Table 1).
Exclusion criteria: patients who un t for surgery; tumors, infections and other factors leading to pathological vertebral fractures, younger than 60 years old.

Surgical approach
All surgeries were performed by the same surgical team to reduce variability related to surgical techniques.
Under general anesthesia, the patient was placed in the prone position with abdomen suspended and connected to neurophysiology monitor. Adjusting the patient's posture to reduce the fracture by extending the involved levels, and the degree of reduction and the remaining kyphosis was evaluated radiographically with a C-arm.
A posterior midline incision was performed, then the lamina and bilateral facet joints were exposed. Pedicle screws (Medtronic, America) were implanted in two vertebrae above and below the affected level respectively, augmented by bone cement (1.5 ml/trajectory) (Tecres, Italy). Ultrasonic bone blade was used to excise the lamina and the facet joints of the side ( with more severe anterior compression) to facilitate placing the angular compressor to lateral-ventral side of the dural sac in order to minimize retraction of the dura and neural elements as much as possible. The opposite upper and lower facet joints were also resected (Smith-Peterson osteotomy, or Ponte osteotomy if necessary) if residual kyphosis required correction. Bilateral underbent rods were connected to screws to correct kyphosis, which could also contribute indirectly to decompression via shifting of the dural sack posteriorly 10,11 . Appropriate amount of normal saline was added as a medium in the surgical eld, and the ultrasonic probe (1202, Brüel & Kjaer, Denmark) was used to detect the sagittal and axial sections of the spinal cord, thereby determining the location and degree of compression from the ventral side of the dural sac. The facet joint on the side with more severe symptoms was removed. Then under real-time ultrasonography guidance, the angular compressor was placed ventral to the thecal sac through the safe area on the cephalad or caudal side 12 . Retropulsed bone fragment which compressed neural elements were carefully and gently compacted ventrally. During the procedure, neurophysiology monitoring was closely observed for any signal changes,including SSEP and MEP. After con rming no compression of the neural elements with ultrasonography, both rods were temporarily removed. The cleft in the vertebral body was lled with bone cement via the pedicle of the injured vertebrae, followed by implantation of two polyaxial pedicle screws.
The rods were reinserted, and screws properly compressed to correct the kyphotic deformity. Ultrasonography was used again to con rm adequate decompression, and correction con rmed radiographically via C-Arm before tightening the nuts to complete the xation. Lamina and facet joints of xed segments were decorticated for bone grafting, autologous bone particles and allogeneic bone chips were used. Drainage was placed and incision was closed in common fashion.
Postoperative antibiotics were used routinely for 24 hours to prevent surgical site infections, supplemented by regular antiosteoporosis treatment; drainage was removed when drainage volume was less than 50 ml/day, and patients were instructed to gradually stand and walk with custom-made braces after 3 days.
Typical routine follows up at our institution include appointments after 3, 6, 12, 18 and 24 months. Generally,the patients were recommended to take CT scans after 6 and 12 months, and X-rays after 3, 6, 12, 18 and 24 months.

Clinical outcome and statistical analysis
The patient's preoperative and postoperative JOA score, ODI score, local kyphosis Cobb angle, operation time, intraoperative blood loss, and postoperative complications were recorded in detail.
SPSS 22.0 statistical software was used for analysis. Continuous variables were expressed as x̄ ± SD. The variance analysis was used for comparison before and after the operation, and the paired sample t test was used for pairwise comparison.
P<0.05 was considered statistically signi cant.

Results
4 males and 7 females were included in the study. Eight cases had minor trauma. The patients were all accompanied by obvious lower back pain, including 8 patients with numbness and weakness in the lower extremities, 5 patients mainly presented with intermittent claudication, and 2 patients developed perineal numbness and sphincter dysfunction.
Among them, the distribution of affected segments was 1 case at T11 level, 1 case at T12 level, 6 cases at L1 level, 3 cases at L2 level. The Preoperative Cobb angle was 34.3 °±5.4 °, the preoperative Lumbar JOA score (Japanese Orthopedic Association Scores)12.8 ± 2.1, and the preoperative ODI score was 71.4 ± 6.4.
There was no leakage of bone cement into the spinal canal in our case series, and all the pedicle screws were augmented. 4 cases performed Smith-Peterson osteotomy, and 2 cases chose Ponte osteotomy to correct kyphosis.
While no patient showed surgical site infection postoperatively. And one patient suffered from postoperative cerebrospinal uid leakage. After being laid in Trendelenburg position, and removal of the drain, no obvious symptoms related to low intracranial pressure such as headache appeared, and the incision healed successfully.
While all other intraoperative monitoring results were stable, one patient's MEP amplitude decreased about 40% during operation, the amplitude gradually recovered after the ventral decompression of the thecal sac, and there was no obvious neurological deterioration after surgery. All 11 patients had satisfactory outcomes during follow-up, with a median follow-up time 17.4 months (12-24 months). At the time of nal follow-up, the local Cobb angle (34.3 °±5.4 °, 11.0 °±4.6 °, respectively), JOA score (12.8 ± 2.1, 22.1 ± 1.9, respectively) and ODI scores (71.4 ± 6.4, 25.5 ± 5.2, respectively) signi cantly improved compared to the preoperative baseline. (P < 0.05, Table 1). No patient experienced failure of instrumentation such as screw pulling out or breakage. Additionally, no deterioration of neurological symptoms was recorded.

Representative case presentation
A 72-year-old woman was diagnosed as L1 stage III KD and incomplete paralysis at our hospital. The Patient's lumbar JOA score was 12 points and T value was − 3.5. She was treated with "compressing the posterior wall of the vertebral body" and guided by ultrasonography, combined with vertebroplasty and cement augmented pedicle screw internal xation.
She recovered uneventfully and was followed up for 2 years. At nal follow up, her low back pain and intermittent claudication disappeared. In addition, she only had slight numbness in her feet and walked very well and was pretty satis ed with her surgery. body and upper and lower intervertebral discs were not resected, which could avoid destroying the anterior and middle column structure and contribute to preserve the stability of the spine. Meanwhile, the bone cement placed into the injured vertebral body to provide anterior and middle column support, and screws were implanted before the bone cement hardened, so that the bone cement block and internal xation integrated together to form a single load bearing frame. Therefore, the stability of the spine was restored to the maximum extent which may explain why no internal xation failure was observed in our study. Additionally, squeezing and compacting the fragment at the posterior vertebral body by the compressor may theoretically close the vertebral basal vein foramen, reducing bone cement leakage into the spinal canal, and improving the safety of the operation 22 . There was no leakage into the spinal canal in our case series, which might illustrate the effectiveness of this technique in reducing leakage of bone cement. Through the application of ultrasound guidance, the location of the surgical instruments and the dura mater could be observed in real time, and compacting the retropulsed posterior vertebral wall could be achieved directly, thoroughly and accurately, minimizing the irritation of neural elements. Only one patient in this group had transient MEP signal changes during intraoperative neurophysiology monitoring, but the patient did not experience neurological deterioration after surgery, and neurological function gradually recovered during follow-up, suggesting the safety and effectiveness of this procedure. In addition, we designed special thin and hard compressor (angular compressor), minimizing encroachment into the spinal canal as much as possible, and it would not be deformed when squeezing and compressing the cancellous bone which also minimized irritating the dural sac. A delicate stripping instrument was employed to release the possible adhesion between dura and bone fragment, so to reduce the risk of a dural tear. Only one case in this group experienced a dural tear, which was lower than the incidence of previous studies 23,24 . Various degrees (90 °, 120 °, etc.)

Discussion
are designed at the tip of the compressor, which was convenient for compressing and compacting the bone fragment. The operative time was relatively shorter, and bleeding and trauma were reduced and fewer neurological complications occurred compared to corpectomy for decompressive surgery. In this study, the average blood loss was 202 ± 40 ml (150 ~ 275 ml), which was less than that of Zhang et al. 2  From our experience gained during this study, the following are our recommendations for the procedure: Firstly, if there are bridging osteophytes in front of the injured vertebrae with adjacent levels, compression between the pedicle screws of this segment should be avoided to prevent screws loosening. The articular process osteotomy at the other segments was recommended if the obvious kyphosis must be corrected. Secondly, if the residual kyphosis was acceptable or the lesion was located in the lumbar segments, facet joints should not be removed, thus reducing the destruction of the posterior column structure and preserve the bone grafting bed, especially when at the lumbar segments or thoracic ones where ultrasonography indicated the presence of the space between spinal cord and pathological compression when dural sac pulsation before decompression 27 . Thirdly, we recommend performing kyphosis correction rst to indirectly release the nerve tissue before decompression, which can improve the safety of surgery 28,29 . Fourth, when putting the compressor to the ventral side of dural sac, we recommend placing it at the adjacent intervertebral disc level rst, then add saline for ultrasound visualization. The location of the compressor can be seen under ultrasonography to avoid interfering with the nerve tissue during placement.
Fifth, the positional relationship between the nerve tissue and the device must be clearly observed under real time ultrasonography during decompression, and the device must not cause compression to the neural elements. Lastly, the Lshaped angular compressor should rst be placed into the cephalic (or caudal) end of the apex of the encroaching bone fragment. After compressing part of the bone fragment, gently and gradually move towards the highest point. To avoid nerve injury, the soft cancellous bone could be compressed rst to form a posterior bone culvert, then use a thin osteotome to cut the bone edge of the culvert from both margins outside the dural sac sides, and then press the cortex to the ventral side to complete the decompression.
As mentioned above, our study avoids complications related to anterior surgery approach and severe damage to spinal stability due to corpectomy surgery. Therefore, spinal stability was preserved in all patients, and the rate of instrumentation failure was signi cantly reduced. All patients in this group achieved bone fusion at last follow-up. Only one case had cerebrospinal uid leakage, but it did not affect wound healing and neurological recovery. These indicates that this surgical procedure was safe and effective in our study population, but a larger number of cases and long-term follow-ups are still needed to con rm our ndings.

Limitations of this study
This study also has certain limitations: the number of cases is small, and the follow-up time for some cases is short, especially considering the late onset of some complications like adjacent segment disease or delayed hardware failure.
Because the incidence of stage III KD is relatively low, the analysis results of 11 patients are still valuable. This study is a retrospective study with a therefore impaired level of evidence.

Conclusion
Spinal canal decompression by "compressing the posterior wall of vertebral body" with intraoperative real time ultrasonography guidance, supplemented with pedicle screw internal xation and fusion for the treatment of stage III KD, could effectively relieve neural compression and symptoms, signi cantly shorten operative time, and reduce intraoperative bleeding compared with the previously reported data. We therefore consider it a feasible, effective and safe procedure in patients suffering from fractures of stage III KD. Informed consent for publication of photographs and data was obtained from all subjects.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.