The natural history of congenital scoliosis and kyphosis has been well
documented1-3.
The degree of scoliosis produced by a hemivertebra depends on the type, site,
and number of hemivertebrae and the patient's age. Thoracolumbar and
lumbosacral junctions are transitional areas between the mobile lumbar spine
and the less mobile thoracic spine or sacrum. Hemivertebrae located in these
two transitional areas lead to trunk shift. In the thoracolumbar and lumbar
spine, progressive kyphosis may also occur. A lumbosacral hemivertebra,
however, usually does not lead to progressive kyphosis.
A single lumbar hemivertebra (between L2 and L4) can be expected to cause
progression of scoliosis at a rate of 1.7° per year if it is fully
segmented and 1° per year if it is partially
segmented4. When
scoliosis progression of =5° is observed on serial radiographs of our
patients, operative intervention is advised.
The aim of this study was to evaluate the results of lumbar hemivertebra
resection and short-segment fusion through a combined posterior and anterior
approach in a consecutive series of twenty-one patients.
Series of Patients (Table
I)
From March 1987 to July 2002, twenty-one lumbar hemivertebrae were excised
in twenty-one consecutive patients (eight girls and thirteen boys). These
hemivertebrae were located at the L2-L3 level in nine patients, the L3-L4
level in four patients, and the L4-L5 level in eight patients. The mean age of
the patients at the time of surgery was 3.3 years (range, twelve months to
10.2 years). All deformities were congenital scoliosis or kyphoscoliosis. All
but two patients had evidence of curve progression of =5° between two
successive radiographs, and all had a fully segmented or a semi-segmented
hemivertebra. The two patients who did not demonstrate progression had a fully
segmented hemivertebra and had 34° and 37° of scoliosis at the ages of
twelve and fifteen months, respectively.
Anatomical Data
The data on the congenital spine abnormalities and associated conditions
are summarized in Table I. Each
complete vertebra was identified as either thoracic or lumbar. An asymmetrical
transitional vertebra with a costotransverse joint and a contralateral
lumbar-type transverse process was considered to be thoracic, but the absence
of rib was noted. Each hemivertebra was not numbered by level but according to
the two adjacent complete
vertebrae5. For
example, an L2-L3 hemivertebra corresponds to a hemivertebra located between
the second and the third lumbar-type vertebrae. There were five right-sided
and sixteen left-sided hemivertebrae. Eleven were fully segmented and ten were
semi-segmented. Sixteen patients (76%) had other spine abnormalities
(Table I).
Preoperative Evaluation
Preoperative radiographic imaging included standing posteroanterior and
lateral view radiographs of the full spine. All but three patients underwent
preoperative magnetic resonance imaging of the spine to assess the anatomy of
the spinal cord and the spinal canal and to study the segmentation of the
hemivertebra and the growth plate; preoperative myelography was performed for
the other three patients (seen in the beginning of the series). All children
underwent renal ultrasound preoperatively to identify and assess associated
congenital abnormalities of the genitourinary system.
Surgery
All of the operations were performed by the senior author (G.B.). The
surgical technique involved complete excision of the hemivertebra through a
combined anterior and posterior approach.
In the first three patients, the anterior approach was performed before the
posterior approach. In the other eighteen patients, a posterior approach was
performed first, with the patient in the prone position. The posterior
elements of only the convex side of the scoliosis were exposed, and the
posterior part of the hemivertebra, including the pedicle, was excised. The
insertion sites for laminar hooks in the adjacent cephalad and caudad laminae
were prepared, and the posterior wound was temporarily closed with sutures.
Next, the patient was placed in a lateral decubitus position, with the concave
side of the scoliosis facing downward. A separate anterolateral incision was
made, allowing a retroperitoneal approach to the lumbar spine. With
appropriate patient positioning, both the anterior and the posterior incision
could be accessed simultaneously. After complete anterior excision of the
hemivertebral body, a posterior convex compression was performed through the
posterior incision with use of pediatric Cotrel-Dubousset instrumentation in
twelve patients, mini-Harrington instrumentation in six patients, and
sublaminar cerclage wiring in three patients. As the hemivertebrae are in a
more posterior location, the compression forces applied posteriorly permitted
closure of the posterior gap that was created by the resection of the
hemivertebra and also permitted correction of the kyphosis. Grafting with
autogenous anterior and posterior fibular bone was then performed. The fibular
graft was placed anteriorly; it was not morselized and was used as a strut
between the two adjacent vertebral bodies to prevent later kyphosis deformity.
A Cell Saver (Haemonetics, Braintree, Massachusetts) was used during the
surgical procedure, and the patients received a transfusion of recycled blood
at the end of the operation. No wake-up test or evoked potential monitoring
was used intraoperatively. The two approaches were done during the same
anesthesia session for twenty patients and were separated by one week for one
patient as a result of excessive bleeding.
Homologous blood transfusion was necessary for only two patients.
Postoperatively, a premolded rigid brace was used to protect the
instrumentation for six months until spinal fusion was present.
Postoperative Assessment
Standing posteroanterior and lateral radiographs of the full spine were
evaluated postoperatively just before the patient was discharged and at the
time of the most recent follow-up. All radiographs were measured with use of
the Cobb method by an independent observer (P.-L.D.) from another hospital.
The curves measured in the coronal plane
(Fig. 1) were the segmental
scoliosis curve, the total main scoliosis curve, the compensatory cranial
curve, and the compensatory caudal curve. The segmental scoliosis curve was
measured between the two vertebrae immediately adjacent to the hemivertebra,
whereas the total main scoliosis curve was the maximum scoliosis angle
(between the two most tilted vertebrae). Two different measurements of trunk
shift—gravity trunk shift and true trunk shift
(Fig. 1)—were evaluated
on the coronal view. The gravity trunk shift is the distance between a
vertical line (plumb line) drawn from the middle of the T1 body and the middle
of the sacrum. The true trunk shift is the distance between the middle of the
sacrum and a line drawn from the middle of the T1 body and perpendicular to
the biiliac line. The pelvic width is the distance between the two points of
the iliac crests tangential to the biiliac line
(Fig. 1). The gravity trunk
shift and true trunk shift were related to the pelvic width and were expressed
as a percentage of the pelvic width to avoid errors due to radiographic
enlargement.
Segmental kyphosis and global lordosis were measured on the lateral
radiograph. The segmental kyphosis was measured between the two vertebrae
adjacent to the hemivertebra, and the global lordosis was measured between the
superior end plate of L1 and the superior end plate of S1. A kyphotic curve is
expressed as a positive angle value, whereas a lordotic curve corresponds to a
negative value.
Statistical Analysis
The results were analyzed statistically with use of the paired Student t
test, with the level of significance set at p = 0.05.
Associated congenital abnormalities were found in nine patients (43%) and
are summarized in Table I.
There were intrathecal abnormalities in three patients (14%): one tethered
spinal cord, one syringomyelia, and one combined tethered spinal cord and
syringomyelia. Genitourinary abnormalities were found in six patients (29%).
No child had delayed psychomotor development. No patient had a neurological
deficit prior to surgery.
The mean duration of surgery was 300 minutes (range, 240 to 390 minutes)
from the time of the initial skin incision to the end of the skin closure. The
mean drop in the hemoglobin level between the day before and the day after
surgery was 3.2 g/dL (32 g/L) (range, 1.9 to 5.9 g/dL [19 to 59 g/L]),
excluding the values from the two patients who received transfusions.
The average follow-up period was 8.6 years (range, two to 17.4 years)
(Fig. 2). At the time of the
latest follow-up assessment, the mean age was 11.8 years (range, six to 18.4
years). Skeletal maturity was graded according to the Risser sign: four
patients were skeletally mature with a Risser sign of 4 or 5; three patients
had a Risser sign of 1, 2, or 3; and the remaining patients had a Risser sign
of 0.
In the coronal plane, the mean segmental scoliosis curve was 32.9°
before surgery, 11.2° immediately postoperatively, and 9.4° at the
time of the latest follow-up assessment
(Table II). The total main
curve was 34.1°, 13.0°, and 12.3° at the same time periods,
respectively. This represents a mean 71.4% improvement for the segmental curve
and a mean 63.9% improvement for the total main curve, with these differences
being significant (p < 0.001) (Table
II). Gravity trunk shift improved from 11% before surgery to 8% at
the latest followup, and true trunk shift, from 19% to 9% (p = 0.007).
In the sagittal view, the mean segmental lordosis was —1.9°
preoperatively, —4.4° postoperatively (just before the patient was
discharged), and —5.2° at final assessment
(Table II).
The immediate postoperative period was uncomplicated for eighteen patients,
whereas complications were encountered in three patients. One patient had a
mild radiculopathy after resection of an L2-L3 hemivertebra but completely
recovered. There was one superficial wound infection, which healed after a
course of intravenous antibiotics. Acute renal insufficiency due to a ureteral
stone associated with a single kidney developed in one patient, who required
ureteral catheterization.
Eighteen patients required no later additional surgery. In one patient, the
spinal implants were removed because of rod fracture. At the time of rod
revision, the fusion was inspected and appeared to be healed, and no
additional bone-grafting was performed. One patient had hook displacement that
required hook reinsertion. One patient had a pseudarthrosis that required
bone-grafting and revision of posterior instrumentation to treat a progressive
kyphosis.
In the present study, intrathecal abnormalities were found in 14% of the
patients, a rate similar to the rate of 18% found by
McMaster6. With the
use of ultrasound, we found genitourinary abnormalities in 29% of children, a
rate also similar to those previously reported in the
literature7-9.
The primary goal in the treatment of congenital scoliosis due to a
hemivertebra is to prevent the development of a severe deformity that might
require a more extensive corrective procedure. Progression of the scoliosis is
most rapid during the adolescent growth spurt and stops only at the point of
skeletal maturity3.
Because a high percentage of congenital scoliotic curves are progressive and
not responsive to bracing, operative treatment is the mainstay of care. The
three main surgical procedures used to treat congenital scoliosis are fusion
in situ, convex-side growth arrest (epiphysiodesis), and hemivertebra
resection.
Isolated posterior fusion with or without instrumentation is not always
recommended for use in young children because correction is limited and
because the crankshaft phenomenon occurs in 15% of patients of all ages and in
36% of patients who are younger than four years old at the time of
fusion10. In a
series of thirty-one patients who had posterior fusion, Hall et
al.11 reported a
reduction of the average curve from 62° to 40° in eighteen patients
who had Harrington instrumentation and a reduction from 43° to 38° in
thirteen patients without instrumentation. Combined anterior and posterior
fusion adds the potential benefit of greater correction and of sagittal plane
correction because the excision of discs allows greater mobility of the
segments. Combined anterior and posterior fusion also decreases the likelihood
of pseudarthrosis and prevents the crankshaft phenomenon by removing the
growth plates anteriorly. Cheung et
al.12 reported on a
series of six patients with a thoracolumbar hemivertebra treated by convex
fusion combined with concave subcutaneous distraction with instrumentation.
There was a mean decrease in the curve of 41%, from 49° to 29°, at the
time of the latest follow-up evaluation, at an average of 10.8 years.
Convex-side growth arrest, or epiphysiodesis on the convex side, is
appropriate to consider for patients with growth potential remaining on the
concave side. This procedure must be done early (preferably before five years
of age) and before the curve has progressed beyond 50° or
60°13,14.
A search of the English-language literature revealed that 123 patients have
been treated with convex-side
epiphysiodesis13-19.
This technique caused improvement in 48% of patients (range, 20% to 77%),
stopped progression only in 40% (range, 17% to 70%), and allowed curve
progression in 12% (range, 0% to 21%). To obtain the desired effect from
epiphysiodesis, two levels cephalad to and caudad to the hemivertebra have to
be fused if decreasing the curve with growth is the goal of the surgery. In
the lumbar area, this fusion extension may lead to an appreciable loss of
mobility of the lumbar spine. In contrast, our technique of hemivertebra
resection requires fusion at only one level. In a study of patients with
infantile idiopathic scoliosis, not including patients with congenital
scoliosis, Marks et
al.20 concluded
that combined anterior and posterior convex spinal growth arrest alone does
not prevent progression of deformity in this condition and that the addition
of posterior instrumentation can slow or arrest the progression of deformity
but cannot reverse it.
Excision of a hemivertebra was first reported in 1928 by Royle in
Australia21.
Numerous
authors22-36
have reported on series in which hemivertebra resection was performed through
successive or simultaneous anterior and posterior approaches or through the
posterior approach alone. The rates of improvement of the scoliotic curve in
those studies ranged from 24.3% to 71.1%. To our knowledge, no series
comparable to our series of patients with lumbar hemivertebra has been
reported to date.
In their series of seven lumbar hemivertebra resections in which a combined
anterior and posterior approach and fusion without instrumentation were used,
Bradford and
Boachie-Adjei24
reported a mean correction of the scoliosis of 68.1% (from 47° to 15°)
at a mean of 4.6 years postoperatively. A cast or brace was worn for seven to
twelve months. King and
Lowery25 used a
Harrington compression rod (six patients) or no instrumentation (one patient)
after hemivertebra resection through a combined approach, with a mean
correction of the scoliosis of 24.3% (from 37° to 28°). These patients
remained in the supine position for six to twelve weeks postoperatively. Lazar
and Hall29 also
used a single compression rod in eleven patients who had hemivertebra
resection performed through the combined approach. The mean correction of the
scoliosis was 70.2% (from 47° to 14°), but the mean duration of
follow-up was only 2.3 years. Leatherman and
Dickson22 and
Slabaugh et al.23
performed dual Harrington rod instrumentation through the combined approach in
twenty-five and four patients, respectively. The correction averaged 44.2%
(from 77° to 43°) and 35.9% (from 39° to 25°), respectively.
Callahan et al.28
reported a mean correction of 60% (from 40° to 16°) in nine patients
after placement of spinous process wires. Klemme et
al.30 reported a
mean correction of 71.1% (from 38° to 11°) in six patients in whom
large sublaminar suture tapes were used after hemivertebra resection through a
combined approach, but this study had a mean duration of follow-up of only 3.5
years.
The authors of more recent studies have reported on hemivertebra excision
from the posterior approach
only31-34.
Ruf and Harms34
reported on twenty-eight hemivertebra resections in which convex compression
by a screw-rod system was used. At a mean of 3.5 years, the mean correction of
the scoliosis was 71.1% (from 45° to 13°). Complications included two
pedicle fractures, three failures of instrumentation, two additional
operations for curve progression, and one infection. Shono et
al.31 reported on
hemivertebra resection through a single posterior approach in twelve patients,
in whom the mean correction was 63.3% (from 49° to 18°). Nakamura et
al.33 reported on
five patients who had a hemivertebra resection with the use of a Harrington
compression rod alone or with the additional use of a distraction rod. The
mean correction was 54.3% (from 49° to 22°) for three patients with a
thoracolumbar hemivertebra and 32.5% (from 35° to 24°) for two with a
lumbosacral hemivertebra. Loosening of the instrumentation was reported in one
patient in this series.
The use of pedicle screws can be effective in correcting scoliosis but may
lead, in our opinion, to a higher risk of neurological complications and can
be difficult to use in very young children with an open neurocentral
synchondrosis. A case of nerve root compression by a pedicle screw was
reported recently by Ruf et
al.35.
The positioning of the patient has been modified from the beginning of our
series to the present. Over time, we have found it easier to first place the
patient prone to perform the posterior resection of the hemivertebra and to
prepare the placement of the hooks. After these procedures are completed, the
patient is turned to the lateral decubitus position and a separate anterior
approach is done for the anterior resection and anterior bone-grafting. The
posterior wound is used simultaneously, with the patient in this lateral
position, to place the spinal instrumentation to compress the two adjacent
laminae after resection of the hemivertebra. When minimal bleeding is
encountered, the two approaches are performed consecutively during one
anesthetic session. For one patient who had excessive intraoperative bleeding,
the anterior approach was postponed until one week later. The fibular graft
works well as an anterior graft between the two adjacent vertebral bodies to
prevent kyphosis deformity at the site of the hemivertebra resection. The
final lumbar lordosis in our series (—31.3°) was within a normal
range (—31° to —79°) according to the Scoliosis Research
Society37. In our
opinion, fibular grafting is essential to avoid progressive lumbar kyphosis or
hypolordosis. There is a risk of valgus deformity of the ankle associated with
fibular graft harvesting in a growing child, but no ankle valgus was
encountered in this series. However, ongoing monitoring for the development of
donor-side valgus is needed.
The correction was stable and persistent in all but one patient. That
patient, who subsequently required a posterior arthrodesis for the treatment
of progressive kyphosis, was one of the first patients treated in the series,
and the cerclage wire used at that time was not as rigid an implant as the
mini-Harrington or the pediatric Cotrel-Dubousset rods that were used later in
the series. In two patients, substantial scoliosis developed in a more cranial
portion of the spine as a result of another hemivertebra.
The technique that we describe permits good control of trunk shift (gravity
trunk shift as well as true trunk shift) with the need to wear a brace for
only six months postoperatively. The brace is premolded prior to surgery and
is placed on the patient immediately after surgery to protect the area of
fusion spanned by instrumentation.
There was no monitoring of motor or somatosensory evoked potentials in our
series. However, we now routinely use neural monitoring during hemivertebra
resection surgery.
To our knowledge, the present study represents the largest series of
patients with lumbar hemivertebra resection in the medical literature. All
twenty-one patients were operated on by the same surgeon, and all
hemivertebrae were located between L2 and L5. We believe that hemivertebra
resection through a combined anterior and posterior approach is an excellent
procedure for early correction in young children. The treatment corrects the
deformity but also prevents further curve progression from the hemivertebra.
Correction should be performed early, even in very young children, as our
results were stable at the time of follow-up evaluation. The procedure is safe
and, in our series, complications were uncommon. ?