Posterior pedicle screws combined with shortening and release techniques for lumbar and thoracolumbar adolescent idiopathic scoliosis

Abstract

Objective:  To prospectively evaluate the clinical and radiographic effects of posterior surgery with wide posterior shortening release and segmental pedicle screws techniques in a consecutive group of patients with thoracolumbar /lumbar adolescent idiopathic scoliosis.

Methods:  Between April 2002 and July 2005, 114 patients (86 women and 28 men) were enrolled in this study. There were 72 Lenke type 5, 32 Lenke type 6, and 10 Lenke type 3C curves. Radiographic parameters such as coronal plane Cobb angle; lordosis angle; lowest instrumented vertebrae (LIV) angulation; and the distances from the central sacral vertical line (CSVL) to the LIV, to the apical vertebra and to the C7 plumb line, were analyzed. Complication rates were also recorded during follow‐up.

Results:  The average coronal correction was from 61° to 13° (78.6%). In the sagittal plane, lumbar lordosis was normalized from 36° with a wide range (23°–67°) to 42° with a normal range (34°–55°). The LIV had 79% correction of coronal angulations. The center sacral line to LIV was improved from 2.3 cm to 0.5 cm, apex to center sacral line from 5.0 cm to 1.6 cm, and CSVL from 2.7 cm to 0.8 cm. A total of 1460 pedicle screws were placed safely, average 9.6 levels (5–14) were fused. The patients were followed up for an average of 30 months (range, 12–50). There was excellent maintenance of correction at final follow‐up.

Conclusion:  Wide posterior release and segmental pedicle screw instrumentation has excellent radiographic and clinical results with minimal complications.

Introduction

Lumbar and thoracolumbar adolescent idiopathic scoliosis has traditionally been treated with the anterior approach. Dwyer et al. initially advocated the anterior approach in 1969 ¹ . During the past 40 years, anterior instrumentations have been redesigned and the clinical effects have been dramatically improved ² , ³ , ⁴ , ⁵ , ⁶ . However, many complications such as kyphotic effect, pseudoarthrosis, and loss of correction still remained ⁵ , ⁶ , ⁷ . Shufflebarger and Clark have reported good results from treating the lumbar and thoracolumbar curves posteriorly with wide posterior release and hooks instrumentation ⁸ . They also found the combination of this technique with segmental pedicle screws and a 5 mm rod could offer an even better three‐dimensional correction. They documented the excellent clinical results of a prospectively collected series of patient using pedicle screws instrumentation after posterior shortening and release ⁹ .

In order to find out whether this method has the same benefits as the anterior procedure in saving fusion levels, and to avoid anterior approach related complications, we conducted the following study.

Materials and methods

Patient sample

Between April 2002 and July 2005, 114 consecutive patients with a diagnosis of adolescent idiopathic scoliosis (86 females and 28 males) were enrolled in this study. There were 72 Lenke type 5, 32 Lenke type 6, and 10 Lenke type 3C curves. The patients were followed up for an average of 30 months (range, 12–50). The medical records and radiographs of patients were studied during the follow‐up period.

Posterior shortening and release technique

This technique, first described by Shufflebarger and Clark ⁸ , made it possible to release the lumbar motion segment concomitant with posterior instrumentation surgery, and appeared to be less invasive than anterior release. Because many thoracolumbar and lumbar idiopathic scoliosis are hypolordotic in the periapical segments, the posterior shortening procedure is necessary in order to achieve normal sagittal balance. This posterior release and shortening, both outside and inside the canal, is performed at every level, followed by fusion from the inferior to the superior end vertebrae. The structures to be excised in the wide posterior release include the interspinous ligament, the ligamentum flavum, and the facet joint.

Our procedures began with release external to the canal. First, the interspinous ligament and as much of the cephalad spinous process as necessary was removed, then the interspace and inferior aspect of the facet joints were exposed. External excision of the facet joint was accomplished next. Because overgrowth of the inferior articular facet's distal tip often blocks reduction on the convex side of the curve, we excised the distal portion of it as well. After the ligamentum flavum was excised, the inferior medial edge of the superior articular facet was exposed. The anterior ligamentous structures of the facet joint were then excised. Rods could be introduced sequentially into the screws, resulting in correction of the deformity. Alternatively the screws were placed after external joint excision, and the intracanal portion of the release left until after the pedicle screws were placed to minimize ongoing blood loss ⁹ . Screws were placed at every level on the convex side of the curve. On the concave side, screws were placed at the bottom two or three and the top two levels of the curve. Occasionally, screws were placed bilaterally at every level. After completion of the posterior release, enough mobility had been achieved to produce lumbar lordosis. As the pedicle screws were placed, cantilever and compression maneuvers were applied to reduce the distance between screw heads, then the rods were bent with placement and tightening sequences, finally restoring sagittal lordosis and correcting the coronal deformity. Cross‐links were employed at both the cephalic and caudal ends. Extensive decortication for local bone graft was performed using a motorized gouge. Iliac crest grafts were used for fusion (sometimes mixed with artificial bone). Closure was routine with one subcutaneous drain.

Post‐operative patient care

After surgery, all patients were managed in the intensive care unit for one night, then transferred to the regular surgical ward. Postoperative analgesia was patient‐controlled. The drain was removed on the second postoperative day. The patients were mobilized with a brace on the third postoperative day. Heavy physical activity was avoided, and the patient restricted to walking with a brace for the first 12 weeks. At the sixth month, the patients were permitted full activity except contact sports. After one year, all restrictions were removed.

Radiographic data collection

Standing long cassette anteroposterior and lateral radiographs were taken before surgery, one week after surgery, and at intervals of 6, 12, 24, 36, and 48 months. Coronal and sagittal Cobb angles were measured. Other radiographic measurements include bending film to measure the flexibility of the main curve, angle of LIV (lowest instrumented vertebrae) relative to the horizontal, relationship of the CSVL (central sacral vertical line) to the LIV, relationship of the vertical from C7 to the CSVL, and the relationship in distance from the apex to the CSVL.

Results

The mean age of these patients was 15.2 years (range, 10.4–18). No neuromuscular diseases were found in this group. All patients were treated by the senior surgeon, and exclusively instrumented with the pedicle screws system. A total of 1460 pedicle screws were placed, and 9.6 vertebrae (range, 5–14) were fused. The patients were followed up for an average of 30 months (range, 12–50). Excellent maintenance of correction was observed at final follow‐up, with no pseudoarthroses or deep infections (Table 1).

Table 1.

Coronal and sagittal values for the thoracolumbar/lumbar curve of the patients

Parameter	Pre‐surgery	Bending correction	Post‐surgery	Final follow‐up
Coronal Cobb	61° (40°–72°)	32° (52.5%)	13° (0°–30°)	15° (0°–31°)
Lumbar lordosis (L1‐S1)	36° (23°–67°)		42° (34°–55°)	43° (33°–57°)
Thoracic kyphosis (T1–T12)	25° (5°–51°)		28° (22°–40°)	30° (20°–40°)

Results of coronal plane correction of the main curve and the sagittal plane alignment of thoracic and lumbar curves are presented in Table 1. The group had an initial average value of 61° (range, 40°–72°), with bending correction averaging 52.5%. The immediate post‐operative curvature averaged 13° (range, 0°–30°). At final follow‐up, the lumbar curves averaged 15° (range, 0°–31°). Loss of correction was minimal. The sagittal alignment from L₁–S₁ for the thoracolumbar/lumbar curves had an initial magnitude of 36°, with a wide range from 23° to 67°. At final follow‐up, the lordosis averaged 43°, but with a narrow normalized range from 33° to 57°. For the unfused thoracic curves, the preoperative kyphosis was 25° with a wide range from 5° to 51°. Immediately after surgery, the kyphosis was 28°, with a more narrow range from 22° to 40°, and at final follow‐up, the average thoracic kyphosis value increased to 30° (range, 20°–40°). This is presumed to result from normalization of sagittal alignment from fusion of the main curve. Overall, alignment was normalized and loss of correction was minimal (Fig. 1).

Figure 1.

(A–F) 17 years old female, lumbar AIS. Anteroposterior and lateral preoperative radiographs (A, B) demonstrate a significant thoracolumbar curve of 44° from T₉ to L₂, with shift of the C7 plumb off of the center sacral vertical line and 25° angulated and significantly translated distal end vertebrae. Trunk shift is present. Sagittal lumbar lordosis is 47°. The immediately postoperative standing X‐rays (C, D) show centralization and leveling of the lowest instrumented vertebrae, the end of the Cobb measurement, which indicate coronal Cobb decrease to 3°and lordosis increase to 49°. The final follow‐up (40 months after operation) standing X‐rays (E, F) show no pseudoarthrosis, no instrumentation breakage, and the loss of correction is small.

Balance parameters for the main curve are presented in Table 2, and are represented by the improvement in the tilt of the LIV, its centralization in relation to the CSVL, and the distance from curve apex to the CSVL. At final follow‐up, the angle of the LIV had 79% correction, indicating horizontalization of the distal end vertebra. The relationship of the CSVL to the LIV had improved from 2.3 cm displaced to 0.5 cm, reflecting centralization of the distal end vertebra. The distance of the apical vertebra to the CSVL was decreased from 5.0 cm to 1.5 cm. The C7 to the CSVL improved from 2.7 cm displacement to 0.7 cm, reflecting the more balanced position of the trunk. These values demonstrate that the trunks were now centered over the pelvis.

Table 2.

Balance parameters of the patients

Parameter	Pre‐surgery	Post‐surgery	Final follow‐up
Horizontal angle of LIV	26 (16–34)	5.5 (0–12)	5 (0–13)
CSVL to LIV (cm)	2.3 (0.6–3.6)	0.5 (0–1.4)	0.5 (0.1–1.5)
CSVL to Apex (cm)	5.0 (3.0–7.2)	1.5 (0–1.7)	1.6 (0.1–1.8)
C7 plumb to CSVL (cm)	2.7 (0.5–4.9)	0.7 (0–1.6)	0.8 (0–1.9)

CSVL, central sacral vertical line; LIV, lowest instrumented vertebrae.

Complications

There were no pseudoarthroses, neurological complications related to pedicle screws, deep infections, or reoperations. The improved results were attributable to the fact that no canal‐violating implants such as sublaminar wires or laminar hooks were used. The lack of loss of correction was due to the use of screws at every level, providing true segmental fixation, coupled with the wide release, which exposed a large surface area of contact within the facets, possibly permitting an early facet arthrodesis. Complications included 12 stress ulcers and three cases of pancreatitis (all patients recovered after eliminating stress factors, general support, drainage with nasogastric tube, and intravenous injections), two wound infections (resulting in one patient from incisional fat liquefaction and in the other from inadequate drainage, both recovered with irrigation and debridement), and one cerebrospinal fluid (CSF) leakage (a dural tear occurred due to inadequate experience of the posterior release technique in the early stage of this study, the patient recovered with careful repair and postoperative management).

Discussion

All the operations were performed by the senior surgeon at one institution. Excellent follow‐up results are documented here: 78.9% coronal Cobb angle correction immediately after surgery, normalization of the sagittal plane, and excellent correction of the balance parameters. However, there are also several limitations of this technique. A large number of pedicle screws potentially increases the risk of screw related neurological complications and pedicle fractures, and the patients' economic burden. There was no randomized control group, which undermines the significance of the statistics. The screws were placed bilaterally at every level in only some of the patients, which decreased the overall correction rate in this study. The average follow‐up (2.5 years) was too short to determine the true complication rate (pseudoarthrosis, crankshaft phenomenon, trunk imbalance, adjacent disc degeneration, and low back pain). This study did not use an outcome instrument, and there was no measurement of apical segment rotation as a third dimension of the study.

The anterior approach for lumbar scoliosis was initially advocated by Dwyer et al. in 1969, using vertebral body screws and a compression cable ¹ . After that, Zielke combined screws with a rod to improve derotation of the vertebrae ¹⁰ . Many published series have shown that anterior discectomy and anterior instrumentation achieve excellent results in such curves. It has been indicated for lower thoracic, thoracolumbar, lumbar and some double curves. However, loss of correction over time, and a significant pseudoarthrosis rate have been documented in the final follow‐up for both the Dwyer instrumentation ¹¹ , ¹² and the Zielke instrumentation ² , ¹³ , ¹⁴ systems. With improvement of instrumentation and the introduction of solid rods, anterior spinal fusion with three‐dimensional instrumentation has been demonstrated to produce excellent correction of the coronal curvature and sagittal kyphosis capacity, while saving fusion levels as compared to posterior fusion with segmental hooks or sublaminar wire, however, several disadvantages still remain ⁵ , ¹⁵ .

Although the posterior approach is familiar to most spinal surgeons, and offers a safe and extensile approach that exposes the entire vertebral column, this traditional method with hooks and rods has several potential pitfalls (more fusion levels, and low back pain in the long‐term follow‐up, which limits its use in the treatment of thoracolumbar/lumbar scoliosis ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ . Nowadays, third generation segmental instrumentation with the pedicle screw system has provided many advantages over other methods of spinal fixation including: better pull‐out strength ¹⁶ ; greater control in the sagittal, coronal and rotational planes due to increased stability to axial, bending and rotational forces by three‐column fixation ¹⁷ , ¹⁸ ; fewer vertebral motion segments arthrodesed ¹⁹ , ²⁰ , ²¹ ; decreased rates of curve progression and higher fusion rates; avoidance of the morbidity and complications (opening of the chest cavity, employment of chest tubes, and the scar being in the flank) of the anterior approach ²⁰ , ²¹ , ²² ; and direct apical vertebral derotation to enhance correction and potentially obviate the need for a thoracoplasty ²³ .

There is some evidence that anterior techniques have required shorter constructs to manage similar curves ⁷ , ¹⁵ . However, with the use of pedicle screws and the wide posterior shortening and release technique in our study, the LIV is the lower end vertebrae of the Cobb measurement in our study, and we have fused the same levels as for anterior instrumentation.

The technique of wide posterior release has previously been reported, with improved coronal correction and production of lordosis (76% coronal correction rate vs. 64% of the control group), but that was just an intraoperative study using only hooks ⁸ . Two retrospective studies have documented that, with the posterior release technique and hooks instrumentation, coronal correction was 67% in both series, with minimal loss of correction at minimum 2‐year follow‐up ²⁴ , ²⁵ . Barr et al. have reported their results using a combination of hooks and pedicle screws versus only hooks ²⁵ . With a ‘complete facet joint excision’ technique, 72% correction of the lumbar curve in the screw/hook group was seen compared with 60% with hooks alone. Loss of correction was 5% with screws/hooks and 13% with hooks alone. The most recently published series on thoracolumbar and lumbar curves using pedicle screws and a posterior release and shortening technique was reported by Shufflebarger et al. in 2004 ⁹ . It was a prospective clinical research, including 61 patients, with a 2‐year minimum follow‐up. All patients were evaluated clinically and radiographically at intervals up to 36 months. No reoperations for nonunion, infection, or instrumentation failure were documented. With generous removal of bony and ligamentous structures within the neural arch, they reported nearly as good coronal correction as with anterior discectomy and instrumentation, 80% coronal plane correction, and normalization of the sagittal plane and balance parameters. However, the authors included double major curves (Lenke types 3C and 6), where there was less rotational deformity and much less cosmetic deformity. Therefore, compared with other series that report only on single thoracolumbar and lumbar curves, the validity of their results should be questioned. Lenke type 5 patients were enrolled exclusively in our study; the results showed superior correction in the coronal plane (the average correction rate was 78.9%), and excellent maintenance and normalization of lumbar lordosis, which is attributable to the wide posterior release and use of pedicle screws in a segmental manner. This centering ability, coupled with normalization of the sagittal plane, may protect distal levels from degeneration at long‐term follow‐up.

The current study analyzed the radiographic parameters and clinical results of surgical treatment for thoracolumbar and lumbar adolescent idiopathic scoliosis using pedicle screws instrumentation and a posterior shortening and release technique. All patients had saved fusion levels (only the Cobb measurement need be fused), and avoided the complications of the anterior procedure, with 78.9% coronal plane correction and improved balance parameters.