Vertebral column resection (VCR) is a formidable last resort technique in the present surgical armamentarium reserved for the most tenacious spinal deformities that cannot be brought to an acceptable range of correction with less aggressive methods. 3 In rigid deformity with decompensation, translation of spinal column is necessary for restoration of trunk balance as well as deformity correction. However, angular osteotomy does not allow translation of the spine, and vertebral column resection is the only option for accomplishing this function.
Vertebrectomy was first illustrated in 1922 by MacLennan, 10 who demonstrated an apical resection from a posterior-only approach with postoperative casting for the treatment of severe scoliosis. Several authors 4–6,8,9,14,15 have subsequently reported their experiences with vertebrectomy, mostly for congenital scoliosis. In 1987, Bradford 1 presented 16 cases of vertebrectomy in patients with fixed multiplanar deformities. He subsequently performed both anterior and posterior vertebral column resection with spinal shortening and posterior instrumentation and fusion with fixed coronal decompensation. 3 The procedure was more extensive than the vertebrectomy procedure for congenital hemivertebra.
Theoretically very appealing, the vertebral column resection is, in fact, a challenging procedure and is an ordeal for both the patient and the surgeon, requiring an exhaustively lengthy operation with a great risk of major complications at every turn en route.
To mitigate the technical difficulties that can be encountered during the process of vertebral column resection comprising a combination of an anterior and a posterior procedure under same anesthesia or in stages, the authors devised a technique of vertebral column resection performed through a single posterior approach (posterior vertebral column resection, or PVCR) that would reduce the operation time and effort, as well as the complications of lengthy combined procedures.
Materials and Methods
Of 215 spinal deformity patients treated during the period of June 1997–March 1999, 143 patients were treated by posterior osteotomy/vertebral column resection procedure. Excluding the patients who could be considered as undergoing posterior osteotomy (intravertebral wedge resection, closing wedge spinal osteotomy without active anterior column reconstruction), 78 met our definition of those undergoing posterior vertebral column resection (resection of all three columns, unopposed gap in the vertebral column). Of these, 70 patients (34 male and 36 female), who were followed up for more than 2 years after the surgery, were retrospectively reviewed using the clinical records and the radiographic materials. The clinical records were reviewed for demographic data, etiology of the lesion, reason for adopting PVCR for correction, operation time, average blood loss, functional improvement, and complications. The radiographic materials were reviewed for preoperative flexibility measured by passive bending radiographs, deformity correction, spinal balance in the coronal and the sagittal plane, complications related to the instrumentation, and stabilization. Deformity was measured by using the standing anteroposterior and lateral 14” x 17” radiographs based on the Cobb’s method. The coronal spinal balance in the frontal plane was determined on standing or a sitting long cassette radiograph by measuring the C7 plumb line distance from the center sacral line. The sagittal spinal balance was determined by the standing lateral films measuring the C7 plumb line distance from the posterior superior corner of S1. A trunk shift of more than 2 cm in the coronal/sagittal plane was considered imbalance.
Etiologic diagnoses were adult scoliosis in 7, congenital kyphoscoliosis in 38, and postinfectious kyphosis in 25 patients. The mean age at the time of surgery was 27.4 years (range 18–64 years old) with follow-up ranging from 2 to 3.3 years, with an average of 2.8 years.
In the adult scoliosis group, the major curve was more than 80° with flexibility less than 25%. In the congenital kyphoscoliosis group, all patients had trunk imbalance with flexibility less than 25%, and all patients had fully or partially segmented hemivertebra. In the postinfectious kyphosis group, all patients showed rigid kyphosis with fused vertebral body and flexibility less than 10%.
Thirty-one patients had resection at the thoracic level and 39 patients at the lumbar level.
In the adult scoliosis group, the major scoliosis curve was 111 ± 25° (range 80–143°) with flexibility of 18.2 ± 6.6% (range 5.2–24.8%). The trunk imbalance was 3.3 cm (range 2.2–4.5 cm), and shoulder height difference was 15 mm (range 0–42 mm) (Table 1). In the congenital kyphoscoliosis group, the major scoliosis curve was 57 ± 18° (range 20–105°) with flexibility of 20.7 ± 4.9% (range 8.1–25.0%), and the kyphosis was 44 ± 25° (range 8–95°) with flexibility of 16.9 ± 5.8% (range 4.9–24.4%). The trunk imbalance was 3.2 cm (range 2.1–6.5 cm) in the coronal plane and 2.7 cm (range 1.6–5.8 cm) in the sagittal plane. The shoulder height difference was 24 mm (range 0–49 mm) (Table 2). In the postinfectious kyphosis group, the kyphosis was 68 ± 34° (range 30–147°) with flexibility of 7.3 ± 2.2% (range 0.5–9.2%). The sagittal imbalance was 4.3 cm (range 2.7–6.9 cm) (Table 3). All the cases had asymmetry between the length of the convex and concave column of the deformity, resulting in coronal or sagittal imbalance and limited flexibility, and translation of column was necessary for restoration of trunk balance as well as deformity correction.
All the surgeries were carried out by the senior author (S.-I.S.) using the method described below. The resections were performed at the apex of the deformity to increase the effectiveness of the resection. The number of vertebrae removed was determined by the type of the deformity and the desired correction to restore the trunk balance. In the adult scoliosis group, one vertebra resection could correct approximately 40° of scoliois. In the congenital kyphoscoliosis group, all the offending anomalous vertebrae were resected. In the postinfectious kyphosis group, all the fused vertebrae were resected.
In the adult scoliosis group, the deformity was fused from one level above the upper end vertebra of index curve to one level caudal to the lower end vertebra of the index curve. In the congenital kyphoscoliosis group, all the vertebrae in the index curve were fused. In the postinfectious kyphosis group, all the fused vertebrae were resected and fusion was carried out from three vertebrae above the resection to two vertebrae below the resection in most of the patients.
All patients were monitored intraoperatively using somatosensory-evoked potentials (SSEP).
Positioning and Anesthesia.
The patients were placed in prone position on a Jackson table under general anesthesia using intravenous neuroleptics to facilitate intraoperative SSEP monitoring.
Incision and Exposure.
The incision could be a straight posterior midline or curvilinear incision depending on the type and size of the subject deformity. Following a subperiosteal dissection, the vertebrae between the uppermost and the lowest instrumented vertebrae were exposed to the tips of the transverse process. The dissection was then carried out laterally, exposing the ribs corresponding to the level of the vertebral column resection.
The facets included in the fusion levels were destroyed by inferior facetectomy and removal of the articular cartilage to promote intra-articular arthrodesis. For the ankylosed or fused posterior facet joints, no attempt was made to mobilized the joints at this stage.
Pedicle Screw Fixation.
Pedicle screw fixation was carried out using K-wires inserted at the presumed entry points and intraoperative radiograph controls. Putting the pedicle screws before the resection procedure had three functions: 1) to provide reliable intraoperative stability to the vertebral column while the destabilization took place; 2) to offer a grip for the vertebral column for the manipulative correction of the deformity; and 3) to provide a radiograph traceable marker for determining the position and the orientation of the vertebral resection. For optimal correction of the deformity and maintenance of the stability, pedicle screws were inserted segmentally, except for the resected levels. In this case, it was not necessary to insert pedicle screws in every segment, but at least four points of fixation on either side of the vertebral resection should be secured before any attempt at vertebral resection. Following insertion of the pedicle screws, the screws were connected on one side with a rod contoured to the shaped of the deformity without any attempt at correction. In scoliosis, the stabilizing rod was placed into the concave side, as it was easier to start the vertebral column resection from the convex side of the curve.
Resection of the Vertebral Column.
To maximize the effect of the resection, it was carried out at the apex of the deformity. The resection began with a removal of the posterior elements. Following a total laminectomy and bilateral total foraminal unroofing to expose the neural elements, the transverse process and the corresponding rib on the working side of the vertebral column (opposite the side with the temporary rod) were removed to expose the lateral wall of the pedicle. The meticulous subperiosteal dissection was deepened following the lateral wall of the vertebral body until the anterior surface of the vertebral body was comfortably palpable. Under visual control, the pedicles and the lateral portion of the vertebral body were removed by using a small osteotome. In the thoracic spine, the rib heads were removed at this stage to allow complete resection of the lateral wall of the vertebral body and to allow untethered motion of the vertebral column. The vertebral body and the intervening discs were removed in a piecemeal fashion gradually towards the medial side and over to the other half of the vertebral body through the void created in the vertebral body, keeping a thin shell of bony posterior vertebral wall beneath the dural tube. The anterior walls were also removed in a piecemeal fashion, taking care to leave the soft tissue tube anterior to the vertebral bodies intact. Attempts were made to remove as much vertebral body and disc as possible at this stage, even across the midline, as it was safe to work with the posterior wall protecting the neural elements. In the process of resection of the vertebral body and disc, meticulous subperiosteal dissection, avoiding injury to segmental vessels, was performed for exposure of the lateral wall of the vertebral body. If segmental vessels were injured during dissection, the bleeding was controlled by electric cauterization and Surgicel. When an adequate amount of vertebral body was removed, all of the posterior vertebral wall that was visible lateral to the dural tube was removed with an Epstein reverse-cutting curette and pituitary forceps. Following the resection of the posterior wall on the working side, another temporary rod, contoured to the shape of the deformity, was inserted to the working side and was securely locked to the screws. Then the rod on the other side was removed to allow resection on that side. The same procedure was carried out on the opposite side. In resection of thoracic vertebrae, the thoracic nerve root on the working side was cut to facilitate resection of the body and reconstruction of the anterior column, but the opposite-side nerve root was saved. In lumbar vertebrae, the nerve roots on both sides were kept intact. At the completion of the resection, the rod that had been removed was replaced and connected to the screws on both sides. It was followed by the final check that the canal was clear of any residual compression at the resection margins and redundant bony or disc tissue attached to the anterior side of the dura that might hinder free, untethered movement of the dural tube.
Deformity correction was carried out either by in situ rod bending or by exchanging the temporary rods with those precontoured to the desired (corrected) shape one by one, and extension of the operating table was unnecessary. The precontoured rod was advantageous in reducing the operative time and the screw failures from force concentration of a specific screw. To avoid inadvertent distraction of the neural elements, the vertebral column was initially shortened by slight compression over the resected gap without tight locking of temporary rods. The deformity was gradually corrected with the repeated additional compression and shortening of the vertebral column. The compression and shortening over the resected gap was carried out until the exposed cord looked redundant. In the adult scoliosis or kyphoscoliosis groups, the compression and shortening over the resected gap could be asymmetrical even with more compression and shortening of the convex side. After compression and shortening of the resected gap, the temporary rods were changed to precontoured final rods one by one to avoid any loss of shortening of the resected gap. In the adult scoliosis group, the curve was corrected further not only by the derotation but also the cantilever method. In the kyphoscoliosis and postinfectious kyphosis groups, the deformity could be corrected further by in situ rod bending and segmental compression.
Anterior fusion or circumferential fusion across the resection gap was carried out, and posterior fusion was performed at all instrumented levels. For anterior fusion at the resected area, autogenous cancellous chip bone graft or titanium mesh was used. After resection and deformity correction, the height of the anterior interbody gap was measured. If the shortest height was less than 5 mm, autogenous cancellous chip bone was placed into the anterior gap. If the height was more than 5 mm, titanium mesh filled with bone chip was inserted into the anterior gap, and autogenous iliac chip bone was placed around the titanium mesh. Titanium mesh was more convenient than autogenous tricortical strut to readjust the size of the interbody graft several times. In the case of the mesh insertion, the mesh was just fit for the anterior gap, and it did not result in lengthening of the anterior column. The mesh cage was inserted from the posterolateral side, through the space between the nerve roots, to fit on the proximal and distal bone bases. The additional compression over the cage was carried out to lock it into place. Unilateral posterior bridging bone graft over the resection gap was done in the thoracic level for the circumferential fusion in 10 cases; however, it also had a risk of postoperative hematoma formation by hindering the evacuation of the blood that accumulates in the canal following the surgery (2 cases of hematoma evacuation among 10).
After bone graft, three or four closed suction drains were inserted at the resection site, and the surgical wound was closed layer by layer.
The patients were allowed to sit up in bed for 24 hours after the surgery. Patients were allowed out of bed with a body jacket cast at the second postoperative week. The body jacket was kept for 3–4 months and followed by a custom-made plastic thoracic lumbar sacral orthosis for an additional 3 months.
In the adult scoliosis group, 1.4 (range 1–3) vertebral segments were removed with fusion of 9.8 levels (range 6–15). The operating time was 337 minutes with average blood loss of 4820 mL (range 3200–6300). The resection was performed at the thoracic level in eight patients and at the lumbar level in two patients. The most common site of resection was at T8. Anterior reconstruction was with titanium mesh cage in two and with cancellous chip packing in five patients (Table 4).
In the congenital kyphoscoliosis group, 1.1 vertebral segments (range 1–2) were removed with fusion of 4.7 levels (range 2–10). The operating time was 243 minutes with average blood loss of 1450 mL (range 800–2600). The resection was performed at the thoracic level in 18 patients and at the lumbar level in 23 patients. The most common site of resection was at L1. Anterior reconstruction was with titanium mesh cage in 1 patient and with cancellous chip packing in 37 patients (Table 4).
In the postinfectious kyphosis group, 3.7 vertebral segments (range 1–7) were removed with fusion of 7.8 levels (range 5–13). The operating time was 296 minutes with average blood loss of 2980 mL (range 1800–4200). The resection was performed at the thoracic level in 50 patients and at the lumbar level in 42 patients. The most common site of resection was at L2. Anterior reconstruction was with titanium mesh cage in 2 patients and with cancellous chip packing in 23 patients (Table 4).
At final follow-up, fusion appeared solid in the standing anteroposterior and lateral radiographs in all cases. Loss of correction was minimal at the 2-year follow-up (Tables 1–3).
In the adult scoliosis group, preoperative scoliosis of 111 ± 25° was corrected to 50 ± 20° following the surgery (range 23–76°) showing an immediate correction of 56.4%. At the final follow-up, it was 54 ± 22° (range 24–82°) showing a loss of correction of 5.7%. Postoperative trunk deviation was 0.6 cm (range 0–1.3 cm) showing an improvement of 27 mm. Postoperative shoulder height difference was 10 mm (range 0–28 mm) showing an improvement of 5 mm (Figure 1;Table 1).
In the congenital kyphoscoliosis group, the preoperative deformities of 57 ± 18° in the coronal plane and 44 ± 25° in the sagittal plane were corrected to 19 ± 13° (range 4–49°) and 2 ± 20° (range −24–55°), showing an immediate postoperative correction of 67.6% and 42°, respectively. At the final follow-up, the deformity was 22 ± 13° (range 6–52°) in the coronal plane and 4 ± 21° (range −24–57°) in the sagittal plane, showing a loss of correction of 5.6% and 2°, respectively. Preoperative 3.2 cm coronal and 2.7 cm sagittal balances were improved to 0.7 cm (range 0–2.2 cm) and 0.4 cm (range −0.2–1.3 cm), showing an improvement of 25 mm and 23 mm, respectively. The shoulder height difference was changed from 24 mm to 8 mm (range 0–18 mm), showing an improvement of 16 mm (Figure 2;Table 2).
In the postinfectious kyphosis group, the preoperative deformity of 68 ± 34° was corrected to 12 ± 24° (range 0–58°). At the final follow-up, the curve was 15 ± 13° (range 0–62°) with a 3° loss of correction. Sagittal balance changed from 4.3 cm to 0.8 cm (range −0.1–1.7 cm), showing an improvement of 35 mm (Figure 3;Table 3).
Complications were encountered in 24 patients (Table 4). The most serious complications were complete cord injuries in two patients, both of whom had adult scoliosis with Beal’s syndrome and postinfectious kyphosis. Before surgery, both of them had neurologic deficit, graded Frankel D. Preoperative baseline SSEP showed diminished amplitude and delayed latency. The complete cord injuries were detected immediately after the surgery. Both patients were reexplored as soon as the conditions permitted. However, the explorations were negative, not revealing any definite pathology or abnormality that could be blamed for the neurologic compromise.
There were six hematomas with cauda equina syndrome. These were evacuated immediately, and all patients were completely recovered within 6 months.
There were five fixation failures. Three patients needed a prolonged localized cast for 6 months, and the remaining patients needed revision surgery. All the revisions remained stable thereafter and appeared to receive solid fusion. There were four root injuries (two in thoracic, two in lumbar; all incomplete), which recovered without further intervention. Two infections were treated by debridement and drainage without any adverse effect on the final result. There were five postoperative hemopneumothoraxes, which were managed by chest tube insertion without prolonged ventilatory care.
Spinal deformity is a three-dimensional deformity. Decompensation both in the coronal and the sagittal plane leads to specific clinical complaints, pain, neurology, progression of deformity, deranged trunk balance, cardiopulmonary compromise, or interference with activities of daily living. In rigid severe spinal deformities, conventional correction methods, such as posterior correction only or anterior release and posterior instrumentation, are usually unsatisfactory; therefore, a more aggressive approach is necessary. 3
Reconstructive techniques, including osteotomy and anterior or posterior reconstruction for the correction of coronal or sagittal deformity, were well described. Leatherman 8 introduced a two-stage anterior and posterior corrective procedure for congenital spinal deformity. Bradford and Bochie-Adjei 2 also reported a single-stage anterior and posterior resection of hemivertebra and spinal arthrodesis. Heinig 7 advocated a combination of circumspinal decompression and correction of the vertebral column in a single posterior approach. He proposed the eggshell procedure for the correction of kyphosis, which consists of a closing wedge osteotomy.
Vertebral column resection differs from spinal osteotomy in that the former creates an unsupported gap in the vertebral column that needs to be supported by an active additional reconstruction, whereas in the latter, the gaps are closed by apposition of the osteotomy surfaces. 3
Posterior vertebral column resection also differs from the conventional combined anterior and posterior VCR in the usage of the spinal instrumentation, mode of intraoperative deformity correction, and method of spinal reconstruction.
We use the pedicle screws with special reduction screws to bring the vertebral column gradually to the precontoured rods, which were sequentially changed from those with minimal correction to moderate correction and finally to the final desired shape. Segmental pedicle screw fixation was also effective in maintaining the vertebral column height with minimal shortening. In cases of anterior gaps more than 5 mm, the titanium mesh cage was used to provide reliable anterior column reconstruction without excessive shortening. The best size of the anterior structural graft would be one that allows a tight fit of the anterior gap, effectively supporting the load and increasing the chance of bony fusion without causing a compression or undue tension in the neural elements. Titanium meshes also had the advantage of offering additional stability because they were strong enough to be punched in and serrated at both ends so that they could be sunk into the bone beds with compression.
Until 1997, the senior author performed anterior and posterior hemivertebrectomy in congenital scoliosis. Although hemivertebrectomy resulted in a satisfactory deformity correction, it was an extensive procedure and still causes procedural problems, such as long operation time, exhausting for patients and surgeons, and the risk of visceral organ damage. In 1997, the senior author developed a one-stage posterior vertebral column resection combined with pedicle screw instrumentation and anterior column reconstruction for correction of fixed kyphotic deformity. 11 This technique was subsequently performed in rigid scoliosis, such as severe adult scoliosis and congenital kyphoscoliosis. 12,13 In our series, all the cases had asymmetry between the length of the convex and concave column of the deformity, resulting in coronal or sagittal imbalance and limited flexibility, and translation of the column was necessary for restoration of trunk balance and deformity correction. Rigid deformity may be treated with conventional correction methods if there is no asymmetry between the convex and concave deformity, if translation of column is not required, and if the shoulder asymmetry is correctible with correction of major curve. Currently, our recommended indication was fixed truncal translation for rigid spinal deformities with flexibility less than 25% in adult scoliosis and congenital kyphoscoliosis. In rigid or postinfectious kyphosis, patients with fused vertebral body with flexibility less than 10% were considered surgical candidates. In rigid kyphosis or postinfectious group, our surgical candidates were usually patients with regional kyphosis more than 80° in thoracic or 30° in lumbar.
We believe that the PVCR has a number of advantages over anterior–posterior VCR: 1) reduction of the total operative time and the amount of blood loss through a posterior-only, one-stage procedure; 2) less risk of intraoperative mishaps due to the instability by enabling the maintenance of spinal stability throughout the resection and the correction procedure; 3) more reliable reconstruction of the spinal column, enabling an immediate anterior structural support that was not possible in the anterior–posterior procedure; 4) less postoperative morbidity than the combined procedure that would allow patients with less than optimal pulmonary function to undergo the procedure; and 5) making the procedure easier in the cases for which the anterior approach normally would have been very difficult (i.e., severe adhesion due to previous anterior surgery, lumbosacral or cervicothoracic area).
Bradford 3 reported on 24 patients with rigid coronal decompensation who underwent anterior–posterior VCR, spinal shortening, and posterior instrumentation. Coronal and sagittal decompensations were corrected to 82% and 87%, respectively. Preoperative scoliosis averaging 103° was corrected to 49° with an improvement of 52%.
The result of our retrospective study confirms our presumptions. It offered a correction of 61.9% in the coronal plane and 47.5° in the sagittal plane, with 25.3 mm and 27.7 mm restorations of coronal and sagittal imbalance, respectively, which were comparable to the anterior–posterior vertebral column resection. The operating time and the total blood loss were significantly less for the PVCR than the anterior–posterior VCR. 3
There were two permanent spinal cord injuries in our series. They were both in patients with severe thoracic spinal deformity and significant clinical and radiologic evidence of cord compromise before surgery. The PVCR was carried out in these patients because it was considered to be the only method that could offer the chance of adequate decompression of the spinal cord as well as correction. In both patients, there was a severe deformity with internal gibbus at the apex of deformity. Though there is no definitive evidence as to what caused the neurologic compromise, the authors believe it might be related to the blood supply of the thoracic cord and preoperative functional status of the spinal cord that would tolerate little additional compromise.
The PVCR is an effective alternative to the conventional anterior–posterior VCR, offering the advantages of reduced operation time and bleeding, more reliable reconstruction of the vertebral column, and abolishing the complications related to the anterior approach. However, it should be kept in mind that this is a technically demanding and exhausting procedure with possible risks for major complications.
- This was the first report on a technique of vertebral column resection through a single posterior approach and its preliminary results in the treatment of moderate to severe deformities with limited flexibility.
- The total number of resected vertebrae was 143: 76 in thoracic and 67 in lumbar.
- The deformity correction was 61.9% in the coronal plane and 45.2° in the sagittal plane.
- Complications were encountered in 24 patients, including 2 complete cord injuries in preoperative cord compromise patients.
- Posterior vertebral column resection is an effective alternative for moderate to severe deformities with limited flexibility.