Excision of hemivertebrae in the management of congenital scoliosis
involving the thoracic and thoracolumbar spine
Journal of Bone and Joint Surgery, May 2001 by Deviren, V, Berven,
S, Smith, J A, Emami, A, Et al
ABSTRACT
We present a study of ten consecutive patients who underwent excision
of thoracic or thoracolumbar hemivertebrae for either angular
deformity in the coronal plane, or both coronal and sagittal
deformity. Vertebral excision was carried out anteriorly alone in two
patients. Seven patients had undergone previous posterior spinal
fusion. Their mean age at surgery was 13.4 years (6 to 19). The mean
follow-up was 78.5 months (20 to 180). The results were evaluated by
radiological review of the preoperative, postoperative and most
recent follow-up films.
The mean preoperative coronal curve was 78.20 (30 to 115) and was
corrected to 33.9 (7 to 58) postoperatively, a mean correction of
59%. Preoperative coronal decompensation of 35 mm was improved to 11
mm postoperatively. Seven patients had significant coronal
decompensation preoperatively, which was corrected to a physiological
range postoperatively. There were no major complications and no
neurological damage.
We have shown that resection of thoracic and thoracolumbar
hemivertebrae can be performed safely, without undue risk of
neurological compromise, in experienced hands.
J Bone Joint Surg [Br] 2001;83-B:496-500.
The management of spinal deformity caused by a hemivertebra is
controversial. The progression of the deformity is unpredictable and
requires continuous evaluation. The location of the hemivertebra is
an important factor in predicting the need for surgical treatment.
When the lesion is in the lower thoracic or thoracolumbar region
surgical treatment may be required to prevent deterioration of the
curve.I The optimum method, however, for the management of a
hemivertebra at these levels has yet to be determined.
Excision of a hemivertebra is a well-established procedure, although
its use has been largely limited to the management of anomalies of
the lumbar and lumbosacral spine. In a classic description of the
aetiology of scoliosis, MacLennan' described the technique of
resection of a vertebral body through a posterior approach, followed
by immobilisation in a cast. He reported "surprisingly little"
correction, however, because of the rigidity of the retained
posterior elements. Von Lackum and Smith carried out a combined
anterior vertebrectomy and posterior fusion in the management of a
fixed lateral deformity, but concluded that the removal of thoracic
vertebral bodies was impractical because of the risk of haemorrhage
and shock. Wiles 11 reported progressive kyphosis in two patients
after excision of a lumbar hemivertebra. Subsequent discussion
revealed that follow-up of the earlier experience of Compete 12 and
of Von Lackum and Smith9 also demonstrated progressive kyphotic
deformity.
Leatherman and Dickson5 introduced the concept of a two-stage
correction using a closing wedge osteotomy with shortening of the
spinal column. Their results gave a mean correction of 43% at follow-
up, with a transient neurological deficit in two patients.5,6 Holte
et al 3 described hemivertebral excision and wedge resection in 37
patients with congenital scoliosis, but reported eight neurological
complications; six followed excision at LS or S1, one after excision
at T10 and one after excision at T9. Bradford and Boachie-Adjei 2
reported on single-stage, lumbar and lumbosacral hemivertebral
excision in seven patients, aged from one to ten years, with a mean
correction of 64% and no neurological compromise.
In spite of these reports showing effective correction with limited
neurological hazard after excision of hemivertebrae, the technique
has usually been used for lumbar and lumbosacral deformities only.
Excision of a hemivertebra above the lumbosacral junction is
controversial as deformity at this level has less impact on spinal
balance, and the risk of neurological damage has been thought to be
very high, especially above the level of the conus medullaris.
Hemivertebral excision has a potential advantage over alternative
techniques for the surgical management of congenital scoliosis by
addressing the deformity directly and allowing immediate, better
controlled and more predictable correction, particularly for
coronally decompensated patients.4,6,7,10
Our aim is to review the outcome of hemivertebral excision in the
treatment of congenital hermivertebrea of the thoracic and
thoracolumbar spine.
Patients and Methods
From our database we identified all patients with the diagnosis of
congenital spinal deformity and the records of patients who had had
thoracic or thoracolumbar hemivertebral excision were reviewed. There
were ten patients with a follow-up of at least two years. In seven,
the procedure had been carried out for coronal deformity, and in
three for both coronal and sagittal malalignment. The excision had
been performed anteriorly in two patients, and through a combined
approach in the remainder. Before this operation, seven patients had
had posterior spinal fusion (Fig. 1). Operative technique. Either a
standard thoracic or retroperitoneal thoracoabdominal approach was
used according to the level of the hemivertebra. Once the level had
been exposed, the discs above and below were excised as far back as
the posterior longitudinal ligament. The hemivertebra was then
removed with a rongeur and curette, including the base of the
existing single pedicle on the convex side. If the hemivertebra was
located at the thoracic level, the head of the rib which articulated
with the hemivertebra was removed to facilitate exposure and
subsequent closure of the space remaining after hemivertebral
excision. The space was loosely packed with autologous bone graft.
As a rule the remainder of the hemivertebra was excised posteriorly
including the rest of the pedicle. Correction and stabilisation were
carried out posteriorly using segmental instrumentation. The extent
of the fusion was based on the preoperative assessment of the
magnitude and location of the deformity, the rigidity of the curve
and the presence of decompensation.
Hemivertebral resection was carried out through an isolated anterior
approach in two patients (one at Ll and the other at T12). In these
cases, in addition to excision of the vertebral body, the convex
pedicle, transverse process and other bone remnants were removed
entirely through the anterior approach. The posterior elements were
not developed substantially; these patients had had no previous
surgery to the spine. They were stabilised by anterior
instrumentation and fusion only.
The effectiveness of the surgery was evaluated by a review of the
radiographs taken before and after operation and at the most recent
follow-up. Absolute measurements were made of the coronal and
sagittal curves, trunk shift, coronal decompensation, thoracic
kyphosis, and lumbar lordosis. Coronal and sagittal curves were
measured according to Cobb's method.14 Trunk shift was determined by
relating the central point of the trunk to the central point of the
pelvis. Coronal decompensation was defined as displacement of the T1
vertebra by more than 25 nun from the central line of the sacrum.
Sagittal decompensation was defined as displacement of the Ti
vertebra by more than 40 mm from the posterior superior sacral
margin.
In addition to the radiological analysis, inpatient and outpatient
records were reviewed. Data were recorded regarding the age at the
time of surgery, the levels fused, the level of the hemivertebra, the
type and level of instrumentation, estimated blood loss,
complications and any additional surgery.
Results
The mean age of the ten patients at the time of surgery was 13 years
(6 to 19). The mean follow-up was for 78 months (24 to 180). Seven
patients had had previous surgery; five a posterior spinal fusion
without instrumentation, one an anterior and posterior spinal fusion,
and one a posterior fusion with Harrington fixation. Four patients
had excision of two hemivertebrae each (Table I).
The mean size of the coronal curve was 78deg (36 to 115) before
operation, which improved to 34deg (7 to 74) at follow-up, with a
mean correction of 59% (45 to 85). The mean compensatory curve was
28deg preoperatively and 11deg at follow-up, giving a correction of
61%. Balance in the coronal plane improved from a mean offset of 36
mm (0 to 60) before operation to 11 mm (0 to 40) at follow-up. The
mean trunk shift was 35 mm before operation and 9 mm at follow-up.
Balance in both planes improved for all patients except one, in whom
there was imbalance in both planes in the cervicothoracic region
after operation (Table II). This patient underwent further surgery
four years later for progressive deformity.
The mean thoracic kyphosis was 35deg (-25 to 76) before and 42deg (18
to 64) after operation. The mean lumbar lordosis was 65deg (28 to 98)
before operation and 52deg (28 to 70) at follow-up. Three patients
with congenital thoracolumbar kyphosis improved after surgery. The
measurements of 80deg, 50deg and 32deg before operation, improved to
32deg, 18deg and 120, after. Alignment in the sagittal plane was
either maintained or improved in all patients.
Two patients required additional surgery during followup. One had
transpedicular subtraction osteotomy for a fixed cervicothoracic
congenital deformity and the other removal of the internal fixation
because of pain. There were no postoperative neurological
complications and no breakages of implants. All patients achieved
solid fusion at the latest follow-up.
Discussion
When congenital deformity of the spine causes an imbalance of growth,
progression of the deformity is rapid and relentless.1,15-19 The
development of a curve is variable depending on the location of the
deformity and the growth potential of the bony elements involved.
Thoracic and thoracolumbar deformities often have a poor prognosis
and usually require surgical intervention. 2,20 There are four basic
procedures available to the surgeon treating congenital scoliosis;
posterior fusion, combined anterior and posterior fusion, convex
growth arrest (anterior and posterior hemiepiphysiodesis), and
excision of the hemivertebra. 2,3,15,19,21-25
Posterior spinal fusion alone has considerable limitations. The goal
of posterior surgery is stabilisation in order to prevent further
progression rather than correction of the curve. Winter 26 reported
290 patients with congenital scoliosis who had posterior fusion with
or without Harrington instrumentation. Correction was limited to 28%
in those fused without instrumentation and to 36% in those in whom
Harrington implants were used. Instrumented distraction across the
concavity was associated with the risk of paraplegia. Deformation of
the fusion mass because of continued anterior growth, was observed in
40 patients (14%). Hall et al22 reported a mean correction of the
curve of 12% in posterior fusions without instrumentation, improving
to 35% with Harrington instrumentation. Slabaugh et alg compared
hemivertebral excision with posterior fusion in situ for lumbosacral
hemivertebrae and found better correction of the curve in the group
who had excision.
Combined anterior and posterior fusion offers several advantages over
posterior fusion. More substantial correction can be achieved by
discectomies, the potential for a crankshaft effect is eliminated,
and the occurrence of pseudarthrosis is reduced. Since this technique
does not address the wedge deformity directly, the entire measured
curve must be encompassed in the fusion, including normal segments.
Convex epiphysiodesis of the spine was designed to arrest convex
growth while allowing concave growth to correct the deformity. The
surgery must take place when sufficient spinal growth remains,
usually in children less than five years of age.15,21,23,27,28
Concave growth is, however, unpredictable and kyphosis in the region
of the anomaly may develop as growth of the posterior elements
continues. It is necessary to perform convex hemiepiphysiodesis
across the entire measured curve, often including a normal segment
above and below, in order to achieve a satisfactory improvement. The
results of this procedure have been variable and unpredictable.
Roaf29 described unilateral hemiepiphysiodesis in patients with
spinal deformity, and proposed that further growth would correct the
deformity. He achieved correction of more than 200 in 23% of
patients, but less than 10 in 40%. Andrew and Piggott 20 demonstrated
mixed early results in a series of 13 patients treated by convex
epiphysiodesis. Long-term follow-up of 33 patients from the same
centre showed correction of the curve in 23 (70%), with better
results in patients treated at a young age.27 Winter and Moe 24
reported early results in ten children treated by convex
hemiepiphysiodesis, with only two demonstrating significant
correction at follow-up at two years. Long-term follow-up of a
similar group of 13 patients showed arrest of the curve in seven
patients (54%) and improvement of more than 5 in five (38%).28
In contrast to the above techniques, excision of the hemivertebra
addresses the deformity directly and allows reliable correction
immediately. It is well established in the management of lumbosacral
curves, which are responsible for pelvic obliquity, apparent leg-
length discrepancy, and truncal listing. Correction cannot be
achieved reliably by other methods. In the thoracic and thoracolumbar
spine, less imbalance is produced but even so, there is often
considerable cosmetic deformity and continued spinal growth may cause
the curves to progress. Hemivertebral excision allows more complete
correction of the curve in these patients, producing improved
cosmetic results and restoration of balance. Our mean rate of
correction of the major curve in these ten patients was 59%, similar
to previously reported results for hemivertebral excision, and much
superior to the radiological results reported for hemiepiphysiodesis,
anterior and posterior fusion, and posterior fusion alone.
Some authors have questioned the safety of such a procedure in the
thoracic and thoracolumbar spine because of the risk of kyphosis and
neurological deficit above the conus. 16 Our results suggest that
hemivertebral excision involving the thoracic and thoracolumbar spine
is not associated with an increased risk of kyphosis or neurological
complications. In ten consecutive cases of hemivertebral excision did
not encounter permanent neurological deficit or progressive kyphosis.
Based on our experience, the correction and balancing of congenital
thoracic or thoracolumbar curves are more effectively achieved by
resection of the hemivertebra than by alternative treatments for
patients with significant, rigid curves.