Congenital Scoliosis

Prevalence studies of scoliosis and spinal deformity do
not distinguish congenital scoliosis from idiopathic scoliosis, and the
true prevalence of congenital scoliosis is therefore unknown, partly
because so many vertebral malformations go unrecognized. Assessment of
a number of congenital scoliosis operations done at a large center
suggests that as many as 20% of cases serious enough to require surgery
among adolescents and children are congenital in etiology (36).

Advances in fetal ultrasound (4,5,6,37,38,39,40)
have permitted earlier and more comprehensive detection of fetal spinal
anomalies. Families may wish to understand the outlook for the child’s
spinal deformity even at the prenatal stage.

Fetal
screening is still an evolving field, and predictions of postnatal
anomalies by prenatal screening continue to be imperfect. Multiple
vertebral anomalies noted on prenatal screening may prove to have
minimal significance postnatally, whereas apparently localized
vertebral anomalies seen prenatally may result in more generalized
defects and syndromic associations with profound postnatal effects. It
is also difficult to accurately predict the postnatal course on the
basis of prenatal neural tube defects. Caution is urged in
prognosticating on the basis of prenatal ultrasound.

The familial incidence of congenital vertebral anomalies is low (31,41). Winter et al. noted that only 13 of 1215 patients had first or second degree relatives with vertebral defects (34).
It would be potentially valuable to provide genetic counseling to
apprise the family of the exact risk of having a subsequent child with
a similar anomaly, but it is difficult to achieve (42). Wynne-Davies (43)
noted a risk of recurrence of 2% to 3% in siblings when multiple
vertebral defects were present, and multiple occurrences have been
documented in the same family (41). Some anomalies such as Klippel-Feil deformity have a clear risk of recurrence in other siblings (2,44). Purkiss et al. (45)
noted a 20% overall incidence of spinal deformity and 17% incidence of
idiopathic scoliosis in families of patients with congenital scoliosis.
For many families there is a sense of guilt associated with the birth
of a child with a congenital anomaly, because they are under the

mistaken
impression that environmental influences or parental factors have
contributed to the congenital anomaly. Although this view may be
correct for some teratogens (46) and for maternal diabetes (14),
most congenital vertebral defects represent a spontaneous occurrence.
Recent research has identified genetic associations for congenital
vertebral anomalies in laboratory animals. In the mouse model,
connections between the genes regulating somite formation and
segmentation have been associated with genes in the “notch” family (2,47,48,49).
Notch genes have been shown to regulate development of the somite and
vertebral precursors in the mouse, and in humans they are associated
with spondylocostal dysostosis and Alagille syndrome (48,49,50).

Associations between known genetic malformations
(including genetic trisomies) and an increased incidence of congenital
vertebral malformations are also well documented (2,19,25,44,45,51,52,53,54,55,56,57,58,59,60).
A few recognized syndromes are associated with congenital scoliosis,
and a knowledge of the genetic basis of these syndromes may help with
planning and screening for future siblings. The genetic markers
associated with some syndromes, such as spondylocostal dysostosis, have
been identified and may be of significance to families (1,47,48,49).
Continued elucidation of the control mechanisms of somitogenesis will
likely lead to identification of other genetic abnormalities among
patients with congenital vertebral anomalies.

All congenital scoliosis results from a failure of
normal vertebral development in early embryogenesis. Early disruption
of normal somitogenesis is presently agreed to be the source of the
failures of segmentation and formation seen in congenital scoliosis (1). Failure to form normal pairs of somites is reflected in hemivertebrae and wedged vertebrae and in hemimetameric shifts (61),
whereas failure in the temporal sequence of somitogenesis from cranial
to caudal manifests as segmentation failures such as block vertebrae
and multiple contiguous vertebral anomalies. Butterfly vertebrae are
presumed to originate from failure of anterior midline fusion between
somite pairs. Use of animal models and inspection of human embryos have
determined that somitogenesis occurs between the 5th and 7th week of
gestation. This is also the time of organogenesis for many organ
systems, and therefore it is not surprising that there is a
considerable association between malformations in the vertebral column
and malformations in the auditory, renal, cardiac, and visceral
systems. Signaling and formation of somite precursors for rib and limb
development occur during a similar interval,

hence rib and limb anomalies are commonly found in association with congenital vertebral anomalies (Fig. 19.2 A,B,C).

The association between vertebral anomalies and teratogens such as carbamazine and valproic acid has long been noted (62),
although the mechanism of action of these teratogens has not yet been
determined. Likewise, the association between maternal diabetes
(particularly type I diabetes) and fetal vertebral anomalies has been
well documented (63).

Transient hypoxia has been shown to produce congenital vertebral anomalies in animal model gestation (10,64,65). Likewise, transient exposure to various toxic elements during the fetal period has produced vertebral anomalies (66);
in these experiments vertebral anomalies and associated skeletal
defects have been similar to those seen in more extensive instances of
congenital scoliosis associated with anomalies of the ribs and
extremities. These same animal experiments have demonstrated that the
location and extent of skeletal malformations are partly dependent upon
the timing of the toxic insult, presumably because of the temporal
sequence of creation of somites from cranial to caudal. Oxygen
deficiency and carbon monoxide exposure, studied in mice, produce
considerable skeletal defects including vertebral malformations.
Although there is no direct evidence linking hypoxia or carbon monoxide
exposure in humans to congenital vertebral defects, there is an
increased incidence of Klippel-Feil syndrome noted among patients with
fetal alcohol syndrome (30,32,67).
Hyperthermia may also be a cause of vertebral malformation and has been
shown to disrupt normal vertebral development in animal models (12,68).

Classification of congenital scoliosis is generally done by modifications of the Nasca classification (74,75). This classification system is based on failure of formation, failure of segmentation, or combinations of the two (Fig. 19.1), and is helpful in describing anomalies and predicting which of them will be progressive (8,74,75,76,77,78,79).
Sagittal plane anomalies are often important concomitant deformities
that should be included in any description. If the failure of formation
or of segmentation occurs only on the right or left side, a scoliosis
results. If the failure occurs only anteriorly or posteriorly, then a
congenital kyphosis or congenital lordosis will occur. Most cases are
subtle combinations of both coronal and sagittal plane deformities, and
a more three-dimensional classification system is needed. Recognizable
patterns of failures of formation include wedge vertebrae and
hemivertebrae in congenital scoliosis, and type I (Winter) failures of
formation in congenital kyphosis (80,81) (Fig. 19.5). Wedged vertebrae result from a unilateral partial failure of

formation, and usually a pedicle is present bilaterally. Unilateral
complete failure of formation manifests as a hemivertebra, recognizable
by the absence of a contralateral paired pedicle. Hemivertebrae can be
subclassified as fully segmented, semisegmented, or nonsegmented on the
basis of the presence or absence of a full growth plate and disc
remnants at the cephalad and caudad sides of the hemivertebra (Fig. 19.1). Hemivertebrae can be either paired or unilateral and can be offset with a “hemimetameric shift” (Fig. 19.4). Failures of segmentation can result in block vertebrae (Fig. 19.6),
anterior or posterior unilateral bars, congenital lordosis, or type II
congenital kyphosis. Complete bilateral failures of segmentation
anteriorly result in a block vertebra. Unilateral anterior failure of
segmentation produces an anterior unilateral bar. Partial failure of
segmentation of the anterior portion of the disc and growth plate
results in a type II congenital kyphosis (Fig. 19.7).
Posterior failure of segmentation can be bilateral and symmetric,
producing congenital lordosis, or unilateral, creating a posterior
cartilaginous or bony bar. Complete anterior and posterior failure of
segmentation creates a block vertebra, whereas unilateral anterior and
posterior failure of segmentation causes a unilateral bar.

The standard classification system of failure of
segmentation, failure of formation, and mixed failures of segmentation
and formation assumes that anterior spinal anatomy is similar to
posterior spinal anatomy. Clinical observation and CT scans (82,83)
show that it is not rare for anomalies in the posterior elements to
show no correspondence with the anomalous anterior elements. The
anterior elements, best seen on plain x-ray film or MRI, guide the
overall classification. Posterior elements are better seen by CT scan
and CT reconstructions.

The lack of correspondence between anterior and
posterior anomalies may explain some of the difficulty in accurately
predicting which deformities will be progressive. If, for example, an
anterior fully segmented hemivertebra is associated with a bilateral
posterior failure of segmentation over the same levels, there may be
slow progression rather than rapid progression, or the progression may
occur more into lordosis than into scoliosis, the posterior failure of
segmentation preventing or slowing the tendency of the fully segmented
hemivertebra to progress. The availability of the three-dimensional
reconstruction of CT scans makes it easier to understand and describe
these anomalies. Radiographic findings may also include associated
anomalies (Fig. 19.2). The presence of rib
fusions or of distortion of the thoracic cavity should be noted. The
presence of multiple rib fusions may act as a paraspinal tether
creating thoracogenic scoliosis, or may enhance progressive deformity
because of the vertebral bar that is commonly juxtaposed to rib
fusions. Examination of the underlying vertebral anatomy and an attempt
at classification is helpful for prognosis (Fig. 19.8).
If certain patterns are identified, such as a fully segmented
hemivertebra, the likelihood of rapid progression may be much higher.
If a unilateral bar

is
noticed, there is a great likelihood of continued progression of the
deformity. The combined presence of one or more fully segmented
hemivertebra as well as a cartilaginous bar and/or rib fusions is
highly likely to lead to progressive deformity with growth (7,8,9,74,75,76,79,84,85).

Absolute Cobb angle magnitude is of much less
significance than overall deformity and the change in that deformity
over time with growth. Documentation of radiographic change with growth
is the mainstay of observation and surgical decision making. Comparison
of current with early x-ray films is needed in order to ascertain
whether there has been deformity progression. Radiographic change may
be very gradual, and archiving of original x-ray films is useful for
later comparison and remeasurement of Cobb angles.

Measurement of sequential radiographs is crucial for the
assessment of deformity progression. Unlike idiopathic scoliosis, in
which a standard measure, the Cobb angle, is fairly reproducible within
and between observers, meaningful documentation of progression in
congenital scoliosis

requires
measurement of reference points that are consistent over time. It is
mandatory that congenital scoliosis be measured in a reproducible way
on the basis of reference to prior radiographs. Reference to only the
immediately prior radiograph may miss slow change and produce
cumulative error in measurement. Because change may occur slowly over
many years, families should be urged to archive original and
intermediate films for later comparison. Of necessity, initial
radiographs of an infant’s spine will be taken in the supine position,
but as the child grows older radiographs should be taken with the child
in the upright position. Caution is needed in interpreting a change in
deformity over this period of time if the comparison is between a
supine-position film and an upright-position film (Fig. 19.2 A,C).
Generally, both posteroanterior and lateral films are necessary. Subtle
worsening of congenital kyphosis or congenital lordosis may go
unrecognized if only posteroanterior films are taken.

Measurement of the Cobb angle of curvature in the
coronal and sagittal planes requires the consistent identification of
end points (86,87).
Identification of end points can be done easily in patients with
idiopathic scoliosis, but requires innovation and consistency in those
with congenital scoliosis. Cobb angle measurement of a pure coronal
deformity may be made by using the vertebral body endplates or,
commonly, the pedicles as endpoints. Measurement of a low lumbar spinal
deformity may be more easily achieved by using the tops of the iliac
crest as the lower end point, in essence measuring pelvic obliquity.
The choice of measurement end points must be consistent between films,
and usually it will be necessary to go back and remeasure films,
starting from infancy, with the same end points in order to determine
whether progression has occurred. It is appropriate to measure not only
the anomalous congenital segment but also the entire curve associated
with it. Attention to secondary curves is equally important, and all
curves should be measured at all intervals. A change in a secondary
curve may reflect an unrecognized change in the associated congenital
“primary” curve. A change in the normally segmented part of the spine
without any change in the congenital curve raises the possibility of
the change being caused by spinal cord tethering, and should prompt an
MRI examination of the intraspinal contents for anomalies (Fig. 19.3 A,B) and tethered spinal cord (Fig. 19.3C) if not already done (3,21,88,89,90).
Other indications of spinal deformity, including trunk imbalance,
apical vertebral deviation, occipital or cervical spine imbalance
relative to the pelvis, and appearance of sagittal plane posture and
sagittal plane balance as seen on lateral radiographs, may be of even
greater significance than the Cobb angle. Loder et al. (87)
found that intra- and interobserver error in Cobb angle measurement in
congenital scoliosis was as much as 13 degrees, thereby suggesting that
Cobb angle alone may be a poor marker for documentation of progression
in congenital scoliosis. Lonstein et al. (86),
however, showed that if sequential radiographs were measured and care
was taken to use the same end points, inter- and intraobserver error in
Cobb measurement of congenital scoliosis was similar to or better than
studies of measurement error in idiopathic scoliosis. Their observation
that sequential radiographs and reproducibility of measurement end
points over time enhance accuracy is of the utmost importance, and the
clinician should follow their advice to re-examine multiple prior
radiographs each time that the patient’s new radiographs are measured.

The short-term natural history of congenital scoliosis
during growth is well understood and documented. The natural history in
the short term ranges from anomalies with dire short term neurologic
consequences, such as spinal dysgenesis (55,70,71,101) or type I kyphosis (Fig. 19.5), to those with rapid deformity change during growth, such as fully segmented hemivertebrae opposite a bar (79,85) (Fig. 19.9),
to those which are completely stable through growth and adulthood. If
the congenital spine deformity is progressive, it generally mirrors
growth velocity, changing rapidly during the first 2 years, slowly
during childhood, and then rapidly again during the preadolescent rapid
growth phase (7) (Fig. 19.11).
Changes in deformity during childhood growth reflect anomalous,
asymmetric, or tethered growth, whereas those seen during
preadolescence may reflect these effects and the effects of gravity as
well. Worsening deformity in congenital scoliosis, congenital lordosis,
or congenital kyphosis may occur most rapidly during the first 2 years
of growth. Progression of the deformity depends upon the balance of
cross-sectional growth in the spine and the secondary effects of any
tethering structures. The presence of intact growth plates unbalanced
by similar growth forces on the opposite side of the spine,
particularly if a tether is present, leads to progressive deformity.

McMaster, Winter, and others have documented the vertebral malformations most likely to demonstrate progression (7,74,76,78,79,102),
noting that progression depends on the type of deformity, its location
in the thoracolumbar spine, the length of spine involved, and the
amount of growth remaining (Fig. 19.8). A
unilateral bar opposite a contralateral hemivertebra is most likely to
progress by as much as 10 degrees per year during rapid growth (Fig. 19.9).
Unilateral bars or fully segmented hemivertebrae alone may also
progress rapidly. Wedge and block vertebrae are less likely to show
progression (Figs. 19.4 and 19.6).
Early prophylactic fusion without waiting for progression may be
justified for some deformities with a high likelihood of worsening such
as congenital kyphosis (Fig. 19.7) or a fully segmented hemivertebrae opposite a bony or cartilaginous bar (79,85,103).

Progression may be more rapid in more mobile sections of
the spine. Lumbosacral hemivertebrae and thoracolumbar hemivertebrae
may show more rapid progression of deformity than congenital
deformities contained within the thoracic spine and constrained by the
secondary stability of the chest. Rib fusions may act as a tether,
causing thoracogenic scoliosis that may either enhance or counteract
the tendency of vertebral deformity to progress (104,105) (Fig. 19.10A). During adulthood larger deformities may continue to progress (106,107), presumably due to the forces of gravity and imbalance acting through vertebral remodeling.

The short-term natural history of congenital spinal
deformity during infancy rarely includes instability and neurologic
change. Neurologic loss caused by spinal canal encroachment may occur
in segmental spinal dysgenesis (11,55,70,71) or type I congenital kyphosis (102) (Fig. 19.5),
but is uncommon otherwise in congenital scoliosis unless there is an
associated intraspinal anomaly, stenosis, instability, or a major
kyphotic component in the deformity (92).
Spinal mobility is restricted in areas of marked congenital scoliosis,
and the concentration of motion stress at the junction between stiff,
abnormal congenital spine and mobile, normal spinal segments may
produce junctional degenerative changes, instability, or stenosis (Fig. 19.6).
Adjacent segment degeneration, instability, and late spinal stenosis
are best documented in association with Klippel-Feil syndrome (2), but can also be seen with congenital scoliosis (106).

Many patients with congenital vertebral anomalies suffer
adverse effects in overall growth and health that are independent of
any associated syndrome such as VACTERL or Klippel-Feil. Concomitant
with abnormalities of vertebral shape, there are consistently observed
deficiencies in spine size and height. Goldberg et al. (72)
have brought attention to the short stature associated with congenital
scoliosis, with and without treatment. One of the later consequences of
disordered growth of the thoracic spine is the effect on pulmonary
function (104,108). Campbell, Day, and others (104,108,109,110,111)
have documented the associations of both congenital and early-onset
scoliosis with pulmonary insufficiency. Pehrsson et al. have linked the
diagnosis of early-onset scoliosis with decreased adult survival
because of pulmonary insufficiency (111). It
seems clear that if congenital scoliosis involves the thoracic spine,
and if the thoracic spine is very short or the chest very distorted,
then the likelihood of major pulmonary insufficiency during adulthood
is increased (104). Campbell et al. have
documented the combination of interrelated thoracic spinal deformity
and chest cage distortion leading to compromise in respiratory function
in progressive congenital and other early-onset spinal deformities (104), and has introduced the useful concept of thoracic insufficiency syndrome to describe the interdependence of chest wall, spinal growth, deformity, and lung function (Figs. 19.2 A,B,C, 19.3, 19.10, 19.13, and 19.14).
By definition, thoracic insufficiency syndrome is the inability of the
spine and chest wall to support normal lung growth and function. In the
series studied by Campbell et al. (104),
thoracic insufficiency occurred in patients with fused ribs and
congenital scoliosis as well as in those with extensive spinal and rib
anomalies such as Jarcho-Levin, spondylocostal, and spondylothoracic
dysplasias (104,112).
Their description of thoracic insufficiency syndrome has led to a
heightened awareness of the potential short- and long-term effects of
early spinal deformity and associated chest wall deformity on pulmonary
function. The inherent shortness of the thoracic spine, associated
congenital rib fusions, distortion of the chest shape with progressive
spinal deformity, and “penetration” of the chest by lordotic thoracic
deformities all contribute to thoracic insufficiency by diminishing
thoracic volume and compliance (Fig. 19.14). Deformity of the lumbar spine, or thoracolumbar or lumbar kyphosis, may also create “secondary” thoracic insufficiency (104) by diminishing the space between thorax and pelvis, thereby in terfering with normal diaphragmatic excursion and respiration (16).

Serial measurement of curvature is the basis for
accurate observation in congenital scoliosis. Difficulty with
radiographic measurement in congenital scoliosis has been well
documented, with a high inter-observer variability noted by Loder et
al. (87). In this frequently quoted study,
intraobserver error for progression in congenital scoliosis was as high
as 13 degrees. Facanha-Filho et al. (86)
documented a much smaller, acceptable inter- and intraobserver error
when care was taken to use the same measurement endpoints and expert
measurers were employed. Observation of change is best done by
reexamination and remeasurement of a range of radiographs begun at
infancy and followed up through the most recent radiograph. Choosing
endpoints for Cobb measurement is arbitrary, so the same endpoints,
whether they be pedicles or vertebral endplates, should be used for
each measurement. Archiving of films is crucial, and it may prove
useful to ask families to retain copies of all the films for the
purposes of repeat measurement. Change in congenital scoliosis may be
rapid in infancy and preadolescence, whereas during middle childhood
years the curve change can be very slow or subtle. Frequency of
radiographic observation should reflect prior behavior of the curve and
growth rate. Observation should include the Cobb angle for all curves,
but also measures of trunk balance. Cobb angle itself may be very
differently interpreted for the same curve. If the most congenitally
distorted section of the curve is measured, it may yield a much higher
Cobb angle than would the overall curve including the congenital
deformity. Other elements of trunk balance, such as deviation of C7
with reference to the sacrum, trunk shift, apical vertebral
displacement, and sagittal balance, should also be measured, and the
results may be more important than the Cobb angle to treatment
decisions. Secondary curves involving noncongenitally involved
vertebrae should also be measured and may be the most accurate
reflection of progression. The secondary curves may act as a marker or
indicator of otherwise unrecognized progression in the congenitally
deformed section of the spine. Films in infancy of necessity need to be
in the supine posture. Upright view x-ray films should be taken

when
practical. The interval of observation with x-rays depends upon growth
rate, curve severity, and prior curve behavior. Growth is most rapid
during the first 2 or 3 years of life, and during this time radiography
as frequent as every 3 or 4 months may be appropriate. If a curve has
remained stable for an interval of observation, it is reasonable to
extend the interval between subsequent observations. The need for early
detection of curve progression must be balanced against the theoretical
risks of repetitive diagnostic radiographs in patients with early-onset
scoliosis (113).

There are few indications for active nonoperative
treatment of congenital spine deformities, there being little evidence
of its efficacy. At times treatment of the noncongenital part of the
spinal deformity with spinal orthoses can be effective. If there is
progression of a deformity within the noncongenital part of the spine,
however, the possibility of spinal cord tethering should be
investigated with MRI; likewise, conditions such as thickened filum
terminale, syringomyelia, or intradural lipoma may be responsible for
curve progression, rather than the actual congenital deformity itself.
There is minimal documentation, and that only in limited circumstances,
about the efficacy of bracing in congenital spinal deformity. There is
no evidence for the efficacy of exercises or chiropractic in slowing
the progression of congenital spine deformities. James noted that the
curve progressed in more than 50% of cases of infantile scoliosis
treated with the advocated casts and Milwaukee braces (114).
Winter et al., however, reported that the Milwaukee brace was useful
for compensatory curves above and below a congenital curve, for long
flexible curves containing congenital elements, and for maintaining
trunk and spine balance following localized fusion during growth (115).
The potential negative long-term effect of casting and thoracic-lumbar
sacral orthosis (TLSO) treatment on the shape of the chest has caused
concern but is not well documented. If long-term bracing is utilized in
the child with

congenital
scoliosis, it should be nonrestrictive to the chest, as is the
Milwaukee brace technique, which carefully applies pressure in
localized areas rather than globally over the chest wall. One should be
sure that the nonsurgical treatment is providing some meaningful effect
rather than just taking up time and permitting development of a more
severe deformity (Fig. 19.14A).

Surgical treatment remains the mainstay of treatment in
congenital spinal deformity. Because the deformity associated with
congenital scoliosis is difficult to correct, early treatment is
preferable to late treatment of a deformity that has become severe.
Allowing the progression of a deformity will inevitably result in
greater risks and a more difficult final surgical procedure (Fig. 19.15).

Preoperative screening tests should include an MRI for
assessment of intraspinal anomalies. The MRI should be used not just to
seek the presence or absence of congenital anomalies but to determine
whether a significant spinal stenosis or disc protrusion exists.
Because segment instability and degeneration with secondary stenosis
with hypertrophy of discs and ligamentum flavum may be present adjacent
to congenital cervical, thoracolumbar, or lumbar scoliosis, there may
be a relative stenosis at either end of a block of congenital
deformity. Dynamic flexion/extension radiographs may be helpful.
Awareness of a relative spinal stenosis preoperatively may be crucial
for planning operative fixation devices or osteotomies. Surgical
planning may include CT assessment using two- or three-dimensional
image reconstruction where possible (Figs. 19.2B, 19.3D, 19.13B, and 19.15D). Newton et al. and Hedequist and Emans (82,83)
have demonstrated the efficacy of three-dimensional CT scan in surgical
planning. Although the congenital anomaly may be readily visible on
preoperative plain x-ray film, the posterior anatomical anomalies may
not correspond to the anterior anomalies seen on plain radiographs. If
surgical correction is planned, and particularly if instrumentation is
planned, it is helpful to have adequate three-dimensional imaging.
Initial preoperative screening tests should also include an assessment
of renal anatomy if one has not already been done. If not already
investigated, a cardiac evaluation including cardiac echocardiography
may be helpful in ruling out substantial underlying congenital heart
disease

If an underlying spinal cord anomaly is found,
consideration should be given to correcting the anomaly prior to
surgical correction of the deformity. Simple deformities such as tight
filum terminale (Fig. 19.3C) or a small
intradural lipoma may be effectively dealt with at the time of the
operation, prior to actual deformity correction. An argument can be
made that more major deformities such as diastematomyelia or major
intradural anomalies should be corrected well in advance of a surgical
correction of the deformity itself (17,21,88,93,94)
in order to prevent the possibly additive adverse effects of deformity
surgery and intradural deformity. The group led by Transfeldt reported
that 3 of 38 patients (8%) experienced worsened neurologic symptoms
when scoliosis fusion surgery was done without previous syrinx
decompression (94). Detethering well in advance
allows one to distinguish neurologic deficits caused by the detethering
from those caused by the surgical procedure itself. Disadvantages of
detethering as a separate procedure include the need for a repeated
surgery on the same area and resultant increased difficulty of the
second procedure because of operative scarring.

Choosing among surgical options for progressive
deformity is complex. In trying to determine guidelines for how to deal
surgically with a progressive congenital spinal deformity, several
factors may help define an appropriate treatment. It is helpful to
assess how severe the deformity is. If the deformity itself is minor
but progressive, a simple in situ fusion
may suffice. If the deformity is severe and causes secondary distortion
of the remainder of the spine, then deformity correction, with its
increased risk, may be justified. Whether the deformity, and hence the
corrective procedure, involves a long or short section of spine can
influence the choice of treatment. Is there impingement by the
deformity on the intraspinal contents? Has the deformity primarily or
secondarily affected respiratory function (thoracic insufficiency
syndrome), and will the proposed treatment create or worsen thoracic
insufficiency? The decision to simply arrest rather than correct
progressive deformity assumes that the existing deformity is
acceptable. The acceptability of an existing deformity is relative. If
correction is very dangerous, then a more severe deformity can be
accepted. If correction is relatively safe, correction with the goal of
improving the deformity may be justified. Therefore, knowledge of all
techniques and an understanding of how much risk is involved are
necessary. Another factor to consider is the effect of the operation on
overall spinal growth (116). If a large segment of the spine is fused in a small child, the overall height of the child may be adversely affected (116)
and thoracic insufficiency syndrome rendered more likely. The
consequences of any fusion on the stability of adjacent segments must
also be considered. Arthrodesis for progressive deformity will make
more of the spine stiff, possibly creating or worsening functional
stiffness and degeneration or instability of adjacent segments. This
dichotomy between extending fusion and creating more junctional stress
is frequently encountered in Klippel-Feil deformity and
cervico-thoracic

deformity.
Extending the fusion to include more vertebrae may help arrest
progressive deformity or instability, but it also concentrates stresses
on a more localized section of spine. Finally, the dangers inherent in
the operation must be examined. The risks of neurologic loss and
intraoperative bleeding are increased by the need for a large degree of
correction or lengthening of the spinal canal, the presence of spinal
stenosis or neurologic deficits, older age of the patient, the need for
circumferential procedures, and other factors (92).

Hemiepiphyseodesis for congenital scoliosis is an
established technique and has been documented in groups led by Winter,
Lonstein, Lindseth, Dubousset, and Thompson (117,118,119) who report a variable epiphyseodesis effect (120,121,122).
Hemiepiphyseodesis can be done posteriorly, or both anteriorly and
posteriorly. When kyphosis is present hemiepiphyseodesis should be done
posteriorly, and when lordosis is present, anteriorly and posteriorly.
Hemiepiphyseodesis can be done as an “eggshell” transpedicular
procedure (119,123).
The anterior portion of an anterior and posterior hemiepiphyseodesis
can also be done thoracoscopically. Immobilization and passive
correction associated with the epiphyseodesis is a critical part of the
procedure, and is best done with a cast. All surgeons document that a
substantial portion of the correction is achieved at the time of the
initial casting and correction (117,118,119,123,124).
The optimal patient for hemiepiphyseodesis is 5 years or less in age,
with a curve that is moderate (50 degrees or less). Compared with a
localized wedge resection for hemivertebra, hemiepiphyseodesis involves
more of the spine, and to some extent builds in an existing deformity.
Further correction of the curves is slow after the initial cast
correction, and some surgeons report the unpredictability of results
following hemiepiphyseodesis (125). The major advantage of epiphyseodesis is that it is very safe and is unlikely to cause new neurologic problems. Posterior in situ fusion (in effect, posterior hemiepiphyseodesis) works well for early moderate kyphosis (Fig. 19.7).

In situ fusion remains the standard treatment for progressive deformity (17,103,107,126,127). In situ
fusion can be performed anteriorly, posteriorly, or both as dictated by
the direction of the deformity and extent of growth remaining. In a
lordotic deformity an anterior fusion is mandatory. In a

kyphotic
deformity posterior fusion alone may help correct the kyphosis by
acting as a tether while continued anterior growth produces some
correction of the kyphosis. In situ
fusion, as classically described, is performed without instrumentation;
however, the availability of modern down-sized instrumentation makes
their use feasible (128), probably diminishing the incidence of pseudarthrosis and length of time the patient must spend in a cast (129).

Spinal fusion with partial or complete correction of deformity is a commonly selected surgical option (Fig. 19.16).
Correction of congenital vertebral deformity can be much more difficult
than correction of idiopathic deformity. Any correction of a congenital
vertebral anomaly carries with it some serious risk of neurologic
injury (92,128). Where
possible, early arrest of progressive deformity is the preferable
option. The underlying congenital vertebral deformities are unyielding
and, if correction is desired, much more than simple instrumentation
may be required. Bending, traction, or supine view x-ray films (that
remove the effect of gravity) may indicate the stiffness of the
congenital vertebral anomaly. Correction beyond that suggested by
bending position radiographs can be achieved by several means. Releases
through discs and facet joints, as with any severe idiopathic
deformity, will permit more correction only if the congenital deformity
is relatively normally segmented. Osteotomy of anterior vertebral bars
or posterior vertebral fusions may be necessary over areas of block
vertebrae or prior fusions (Figs. 19.15 and 19.17).
A decision as to whether deformity correction is worth the additional
risk must take into account the effects of correction on the remaining
mobile parts of the spine. Other risks associated with osteotomy or
releases include excessive bleeding and direct cord injury. Corpectomy,
vertebrectomy, or vertebral column resection may be necessary for
achieving any improvement of the most deformed sections of the spine (130,131).
Traction alone after release or osteotomy remains an option for slow
correction of deformity. Caution should be exercised in using traction
after vertebrectomy, osteotomy, or extensive release, as the segmental
instability of the spine created by the vertebrectomy or osteotomy may
be severe enough to make traction dangerous. In the case of a
preexisting neurologic deficit, traction (halo-gravity, halo-pelvic, or
halo-femoral) (132,133,134) may be either helpful or disastrous, depending upon the spinal anatomy and cause of the paralysis.

Particular care should be exercised if the deformity involves kyphosis,
because elongation of the spine with traction may draw the compromised
spinal cord more tightly against the kyphotic deformity before the
deformity itself improves (81).

Hemivertebra excision has been well documented (19,135,136,137,138,139,140,141,142,143,144,145).
The ideal indication for hemivertebra excision is an isolated, fully
segmented hemivertebra, at the thoracolumbar junction or below, that
has produced a single deformity (Fig. 19.18).
The rationale for early hemivertebra excision includes correction of
the deformity and presumed prevention of subsequent secondary deformity
in the remaining relatively normal spine. Excision of anterior and
posterior hemivertebra elements or “wedge resection” can be performed
through staged or simultaneous anterior and posterior approaches or
through a posterior approach alone. The choice of approach should
depend upon the experience of the surgeon and the location and
direction of the deformity. Anterior hemivertebrectomy followed by a
separate posterior hemivertebrectomy and posterior instrumentation is
the standard practice. The anterior and posterior incisions can be
performed at the same time if desired, although this makes the
posterior instrumentation more awkward to fit. Positioning of the
patient in a manner to permit this procedure has been described (143).

The utility of simultaneous anterior and posterior
approaches includes the ability to readjust the amount of resection
either anteriorly or posteriorly in order to ensure satisfactory
correction. It is also feasible to perform a single posterior or
posterolateral approach for hemivertebra excision. A posterior-only
approach is easier in kyphotic than in lordotic deformities. When
performed on the thoracic spine, the posterior approach may be combined
with costo-transversectomy for added exposure. Difficulty in
controlling vertebral body bleeding may be encountered in any
hemivertebra excision but is probably greatest in a purely posterior
approach.

Hemivertebrectomy performed from any approach should be
planned so as to include enough of the convex hemivertebra and adjacent
cartilage and disc and to extend far enough toward the concavity so
that complete correction can be achieved. Excising the hemivertebra
just to the midline where the bony hemivertebra may end will usually be
insufficient for substantial curve correction. A wedge resection
including the hemivertebra, hinged on the concave edge of the curve, is
generally necessary. Without extensive correction of the deformity
associated with the hemivertebra, the

major
advantage of early hemivertebra excision is not achieved.
Contraindications to hemivertebrectomy probably include the inability
to maintain the correction with a cast or instrumentation; the presence
of rigid curves above or below the excision, which will cause spine
imbalance; a major intraspinal anomaly in the area; and, probably,
location of the hemivertebra in the cervical spine, where the vertebral
artery complicates excision. Early hemivertebra excision is advocated
because, if allowed to progress, a deformity associated with the
hemivertebra may produce more structural deformity over a longer
section of the spine, eventually requiring longer fusion and more
growth arrest. By doing a hemivertebra excision early, one may restrict
the growth arrest to a short section of spine. The timing of
hemivertebra excision depends upon several factors. The longer the
child remains in an upright position with a severe deformity from a
hemivertebra, the more likely the secondary curves are to become
structural. On the other hand, hemivertebra excision in the very small
child, younger than 1 year, is made more difficult by the need for very
small instrumentation.

The neurocentral synchondrosis closes at 3 to 6 years of age (146,147),
but it is probably safe to perform fusion or instrumentation before
that age without fear of creating spinal stenosis, as the canal
diameter at birth is already approximately two-thirds of the adult
dimension (148). Kim et al. (129)
found no instances of permanent neurologic deficit in wedge resections
done before the age of 5 years; in general, the earlier the procedure
was performed, the less likely neurologic injury was (92,129). Other series of hemivertebra excisions have also indicated that early excision makes neurologic injury less likely (141,144).
Injury may come from direct contusion of cord or roots or may come from
stretching of the concave roots as a new position is achieved. Some
hemivertebrae

are nonprogressive (Figs. 19.4 and 19.12)
and do not produce a deformity. Indications for hemivertebra excision
must therefore include either progressive deformity or an existing
deformity that has a considerable effect on the overall configuration
of the spine above and below. In the absence of specific
contraindications, lumbosacral, lumbar, or thoracolumbar locations are
optimal for hemivertebra excision, whereas deformity correction with
hemivertebra excision in the mid- and upper thoracic spine is limited
by the inflexible thorax, and probably entails greater neurologic risk
because of the presence of thoracic spinal cord at the level of the
operation. For an isolated hemivertebra in the midthoracic spine, in situ fusion or hemiepiphyseodesis may be a better choice than excision.

When a block deformity or previous iatrogenic fusion
exists and contributes to congenital spinal deformity, osteotomy may be
required for achieving correction of deformity and balance (Figs. 19.15 and 19.17).
Generally both anterior and posterior osteotomies and/or release will
be needed, and can be staged or done sequentially. Vertebral column
resection (131) may be done in a similar
fashion. Anterior procedures can be osteotomies at each vertebral level
in the area of deformity or can be restricted to a limited number of
osteotomies if localized correction is planned. Posterior osteotomies
also can be done at multiple levels or locally. Pedicle subtraction
osteotomy has gained currency for treatment of adult spinal deformity (149) and has similar application in children’s congenital deformities. Transpedicular eggshell procedures may be carried out (150)
for previously fused pediatric congenital spine deformities. With any
vertebral osteotomy, the risk of neurologic injury is considerable, and
substantial bleeding can be anticipated. Osteotomy technique is well
described in the literature, but each procedure should include careful
localization and preoperative planning. Three-dimensional CT scans (Fig. 19.15D)
will help rationalize the decisions regarding an osteotomy and help the
surgeon localize the site of osteotomy. Inner edges of osteotomies
should be tapered and undercut so that there is no impingement on the
spinal canal at the time of closure. Correction after osteotomy should
emphasize avoidance of spinal column elongation (Fig. 19.15 F,G).

Immobilization is necessary after surgery for correction
of a deformity. Traditional immobilization has included cast fixation
either with or without instrumentation. Cast immobilization has obvious
disadvantages but is entirely feasible in children. Maintenance of
correction with cast immobilization alone requires attention to details
relating to the cast. Many of the instrumentations used in small
children are not strong enough to permit immediate mobilization, and
therefore backup immobilization of instrumentation may be needed for
several months after the operation. Instrumentations used in fusion
with correction can be of various kinds. Preoperative evaluation of the
congenital vertebral anomaly, usually by CT scan, should include an
assessment of available anatomic anchor points for fixation (82,83).
Pedicles may be very distorted or absent and laminae may be distorted,
oblique, absent, bifid, or very thin. Canal diameters may be
restricted. Individualization of fixation, balancing risk against
stability, is needed. The increasing availability of downsized
instrumentation, either intended specifically for smaller children or
adapted from adult cervical applications, has made instrumentation of
deformity in smaller children more feasible and permitted a reduction
in the use of cast and brace immobilization postoperatively (128).
However, the posterior and especially anterior osseous elements in
small children are much less dense than the equivalent structures in
adults and do not permit the application of corrective forces via
anchor points as they would in adults. Where possible, correction and
instrumentation of congenital scoliosis should not involve elongation
of the spinal column, because the spinal cord in congenital deformity
is likely to tolerate elongation poorly.

Traditional approaches to severe early congenital spinal
deformity have included early fusion. The rationale behind this
approach has been to avoid the development of more severe deformities
that would necessitate more extensive and dangerous procedures later in
childhood or adulthood. The use of early fusion to prevent further
progression of deformity assumes that the loss of growth associated
with early fusion is well tolerated. In general, this is true, and the
principle of early arrest of progressive deformity remains the mainstay
of surgical treatment. Some patients, however, particularly those with
an already shortened spine whose deformity would necessitate fusion of
a long portion of spine at an early age, and especially those with
associated chest wall anomalies, may be better treated by techniques
that permit some continued growth of the abnormal spinal elements while
still controlling the deformity. Expandable posterior rods and
instrumentation without fusion have been employed in children with
infantile and juvenile idiopathic deformities, but also in small series
of those with congenital deformity (132,151,152) (Fig. 19.19).
A major disadvantage of this treatment is the need for repetitive
lengthening procedures, but it may be an appropriate temporary choice
in the very young child with a deformity involving a large portion of
the spine.

The group headed by Campbell has pioneered the use of expansion thoracostomy and a VEPTR (153) for the treatment of thoracic insufficiency syndrome (104) in growing children (Figs. 19.10 and 19.13).
Combined spine and thoracic deformity caused by congenital vertebral
deformities associated with multiple rib fusions is particularly
amenable to the use of this technique during growth, and is the
indication best documented by Campbell et al. (112).
By performing one or more expansion thoracostomies through the
constricted chest wall, either between ribs or through the rib fusion
mass, expansion of the chest and secondary control of the spinal
deformity without fusion is possible. The expandable devices attach rib
to rib, rib to spine, or rib to pelvis and are periodically expanded to
permit growth. Eventual fusion near the end of growth is anticipated,
but by expanding and controlling the shape of the chest while
permitting longitudinal growth of the spine, the chance of thoracic
insufficiency syndrome is minimized. Growth of the spine and growth of
concave bony bars has been demonstrated in association with this
technique (100). It was Campbell who brought
attention to thoracic insufficiency syndrome, pointing out that growth
of the spine and chest are intricately intertwined. Primary chest wall
deformity, such as massive areas of fused ribs, associated with
congenital spinal deformity acts as a powerful deforming force on the
spine. Treatment of the spine alone will often be defeated by the
tethering effect of rib fusions or rib abnormalities. In young
children, a primary

chest
wall deformity usually creates a secondary spine deformity, and
conversely, severe progressive spinal deformity usually creates a chest
deformity. While considering treatment of any early-onset spinal
deformity, one should keep in mind the need to prevent irreversible
chest wall deformity and thoracic insufficiency syndrome (104).

Several observations are at variance with general
practice. In my experience, many congenital anomalies predicted to be
progressive are not progressive or are not progressive until the
preadolescent growth spurt. Although most congenital kyphosis is indeed
progressive, and early posterior in situ
fusion is usually the best choice, some instances of nonprogressive
congenital kyphosis have been observed. Three-dimensional CT scan
visualization of posterior elements may explain why some anomalies are
not progressive; fully segmented hemivertebrae seen on plain x-ray film
may include nonsegmented posterior elements on the convex side of the
curve. Therefore the underlying posterior congenital anomaly acts as
its own “tether” on the convex side, preventing progression. This
phenomenon may also be the reason why occasionally congenital kyphosis
does not progress, with a spontaneous fusion of posterior elements
acting as a tether opposite the congenital kyphosis.

It is also my observation that the hemiepiphyseodesis
effect, at least in my hands, is much less effective than is suggested
by the literature, and that the hemiepiphyseodesis effect of further
correction after cast correction is infrequent in spite of careful cast
correction and immobilization. Hemiepiphyseodesis was developed as a
technique in an era when instrumentation was not available for the
congenital spine deformities in very young children and when
intraspinal imaging of anomalies was less advanced. With the
availability of small pediatric instrumentation (128), it now seems more effective to achieve correction over an area of deformity by instrumentation rather than by casting alone.

Transpedicular eggshell vertebrectomies and/or
transpedi-cular hemiepiphyseodesis have, in my experience, been more
difficult than anticipated and less satisfactory. In very small
children, nearly complete obliteration of the pedicle is necessary in
order to achieve enough access to the endplates to assure growth
arrest. Destruction of the pedicles and further disruption of posterior
elements in some instances seems counterproductive to the goal of
achieving a solid fusion.

Congenital rib fusions are often ignored in the planning
for surgical correction or the execution of epiphyseodesis or fusion.
If major rib fusions exist in the concavity of a curvature, it has been
the author’s observation that the rib fusions will usually continue to
deform the affected spine in spite of in situ
fusion, instrumented fusion, or osteotomy. With the present
availability of expansion thoracostomy, rib fusions should be taken
into consideration in surgical planning.

Congenital scoliosis is a potential problem very early
in growth, and severe instances of congenital scoliosis are rapidly
progressive during infancy and early childhood. As a result, there is a
focus on early progression of congenital scoliosis, and a false sense
of security often results if the deformity has not progressed by
mid-childhood. Families and caregivers alike are falsely assured that
if the deformity is nonprogressive during early childhood, it will
remain nonprogressive through the remainder of growth. Late,
considerable progression during the preadolescent growth spurt is often
unanticipated and can be devastating. Continued observational vigilance
is needed all through growth.

In my experience, early posterior in situ fusion for congenital kyphosis or congenital scoliosis with a major kyphotic

component has been highly successful (Fig. 19.7). Supplementation of posterior in situ
fusion with instrumentation lessens the need for cast immobilization.
The principle of early creation of a posterior tether opposite an
anterior congenital kyphosis with anticipated continued anterior growth
and later kyphosis correction is valid and effective. Because the
incidence of perioperative paralysis associated with congenital
kyphosis is high in late reconstructive procedures, particularly
circumferential procedures, early in situ posterior fusion seems warranted and is easily accomplished.

Likewise, hemivertebra excision or global wedge
resection for localized deformity has been highly successful. If the
deformity involves a short section of spine, and particularly if this
short deformed section causes a compensatory deformity above or below
in a relatively normally structured region of spine, hemivertebra
excision or wedge resection should be strongly considered. Usually
fusion can be restricted to the segment just above and below the
hemivertebra, not involving any normal spine above or below with fusion
or instrumentation. Preoperative planning requires the demonstration by
CT scan of posterior elements suitable for fixation and corresponding
to what is seen in the anterior elements on plain x-ray film. Complete
correction of the deformity is necessary in this procedure, and enough
of the hemivertebra and associated disc must be excised to provide
correction. The excision must traverse to the concave corner of the
wedge in order to provide mobility for correction. Simply excising the
bony hemivertebra results in undercorrection and defeats the goal of
complete correction involving a short segment of spine. The use of
growing rod constructs in congenital scoliosis is poorly documented.

It has been my experience that in situ fusion or in situ
instrumented fusion of a short congenital segment of spine and growing
rod construct treatment of the longer, normally segmented section of
spine has worked well for large deformities involving normally
segmented spinal curves adjacent to congenital curves (Fig. 19.19).
Likewise, very long, relatively mobile sections of congenital spinal
anomaly in young children seem to be amenable to growing rod treatment,
provided the curve is relatively flexible and sufficient anchor points
are present. If a chest wall anomaly is present or there is severe
chest distortion in a young child, and thoracic insufficiency syndrome
is a consideration, then expansion thoracostomy with VEPTR treatment is
my preferred choice (Figs. 19.10 and 19.13).
If there is no considerable thoracic deformity and if the problem is
restricted to the spine, it is my preference to use a growing rod
construct. However, if the chest deformity is advanced, then expansion
thoracostomy, as described by Campbell et al., may be the preferable
way to control spinal deformity while still permitting growth and
improving chest volume and shape. The clearest indication for expansion
thoracostomy and VEPTR, in my opinion, is the combination of congenital
vertebral anomalies over a substantial section of spine and concave
fused ribs. If a spine operation alone is performed, the fused ribs
will continue to tether the spine, causing the deformity to worsen (Fig. 19.10A).
Combined treatment of concave fused ribs and extensive vertebral
anomalies by expansion thoracostomy and an expandable rib-to-spine and
rib-to-rib prosthesis has been highly effective in the experience of
Campbell et al. (112) and also in my practice.
Presumably most patients treated with expansion thoracostomy will need
a definitive spinal fusion at the end of growth.

The current popularity of growth-preserving procedures
such as growing rods and expansion thoracostomy makes the choice
between traditional fusion-based treatment and growth-sparing
procedures more complex. Several factors can be considered in choosing
a treatment: the age of the patient, and therefore the amount of growth
remaining; the length of the spine involved; and the availability of a
satisfactory, traditional, early fusion, hemiepiphyseodesis, or
hemivertebra excision procedure for that area of spine. Essentially, if
the deformity is easily treatable with a localized procedure such as
hemivertebra excision or wedge excision involving a short section of
the spine, then this sort of one-time, traditional, fusion-based
procedure is preferable to repetitive surgeries associated with growing
rod techniques or expansion thoracostomy. However, if the child is very
young and a long section of the spine is involved, early fusion-based
techniques will result in growth arrest over a longer, more significant
section of spine. In such an instance, where a lot of growth remains,
the choice of growing rods or expansion thoracostomy in the VEPTR
procedure may be preferable even though it involves multiple procedures
and multiple hospitalizations. Finally, if major rib fusions are
present, the child is very young, and a long section of thoracic spine
is involved in the curvature, then expansion thoracostomy with VEPTR
procedure is preferable.

In making choices regarding surgical treatment of
congenital scoliosis it is of paramount importance to have specific
goals in mind. Treatment should maximize the length of the spine at the
end of growth and should minimize thoracic deformity; it should leave
as many segments of the spine mobile as is possible, and should place
the patient at as little risk as possible. Hospitalizations and
repetitive procedures should be kept to a minimum. For example, a
procedure such as hemivertebra excision for an isolated thoracolumbar
hemivertebra can be expected to diminish total spine height very
little, and calls for a single, brief hospitalization. A more complex
procedure such as expansion thoracostomy and VEPTR insertion will
require multiple procedures and likely be associated with some surgical
problems or complications, but will maximize the available spine length
at the end of growth. Early in situ fusion
for congenital kyphosis, although it may possibly result in less trunk
height than an anterior release and strut grafting, is a less complex
procedure, exposes the patient to a minimum risk of paralysis, and is
therefore preferable.

In spite of surgical treatment (hemiepiphyseodesis, in situ fusion or fusion with correction), progressive deformity occurs with variable frequency as the patient grows (9,103,107,117,118,119,121,137,154). Deformity can recur within the fusion mass through bending, through the crankshaft phenomenon (155,156,157,158),
or by “adding on” to the curvature above or below the prior surgical
treatment. Recurrent deformity can be addressed by extension of fusion
and instrumentation and osteotomy of prior fusions.

Neurologic loss is commonly associated with surgical
treatment of congenital spine deformity, notably congenital kyphotic
deformity (9,92,159,160,161). Early morbidity reports from the Scoliosis Research Society (160) and other reports (9,92)
note the considerable risk of neurologic complications associated with
surgery of congenital spine deformity, particularly congenital
kyphosis. Neurologic problems are associated with progression of the
deformity (159) as well as with surgical
treatment. Although type I congenital kyphosis may be associated with
paralysis if untreated, it is uncommon for untreated progressive
congenital scoliosis to cause neurologic loss unless there is a very
severe deformity, associated kyphotic deformity, lateral translocation
or dislocation (162), a spinal cord anomaly, or spinal cord tethering.

Several possible etiologies of neurologic injury in
congenital spinal deformity explain the relatively high incidence of
neurologic injury associated with treatment. The final common pathway
for most perioperative neurologic complications is probably vascular
insufficiency of the spinal cord. The combination of an inherently
anomalous or insufficient vascular supply to the spinal cord (Fig. 19.15E)
coupled with corrective forces and change in position or shape of the
spinal canal may result in neurologic compromise more commonly in
surgeries for congenital deformities than in surgical correction of
idiopathic deformities. Apel documented changes in somatosensory evoked
potential (SSEP) monitoring with anterior segmental vessel ligation in
congenital scoliosis (163). This has also been
the author’s experience in several instances. The vascular supply of
the spinal cord associated with congenital scoliosis can be anomalous,
and wherever possible surgical procedures should consider the possible
effect of interruption of spinal cord blood supply during anterior
surgical procedures. Distraction of the spinal cord in an animal model
has been shown to be associated with diminished vascular flow, which
precedes the appearance of neurologic dysfunction (164,165).

A similar mechanism probably occurs in neurologic
injuries associated with correction of congenital spine deformity.
Wherever possible, acute surgical correction should shorten the spinal
column rather than distract the spinal cord. There may be value in
staging corrections or producing the correction slowly in traction in
order to allow monitoring of neurologic function when the patient is
conscious (134). Neurologic injury may also
occur from direct mechanical impingement on the spinal canal.
Translation or angulation of the canal associated with spinal column
correction may result in a new spinal stenosis or angular impingement
on the spinal cord.

The risk of mechanical impingement can be lessened by
preoperative imaging assessment of canal anatomy and dimension, and
adequate control, observation, and decompression of the spinal canal
during corrective maneuvers. Undercutting of lamina and procedures that
shorten the area of deformity may make impingement less likely.
Neurologic injuries associated with surgical treatment of congenital
spinal deformity may occur without obvious cause (preoperative stenosis
or spinal cord anomaly or tethering), and may be delayed (81).

The presence of a preexisting neurologic deficit should
be cause for thorough preoperative evaluation, and probably signifies
an increased risk of postoperative neurologic deficit. Early treatment
is preferred where possible, while the deformity is less severe. Also,
during early childhood the spinal cord appears more resistant to
injury, and there is a lower incidence of neurologic injury (129).
However, surgeons should realize that even with the most cautious
approach, there is still a risk associated with manipulation of the
congenitally anomalous vertebral column and spinal cord, and that the
incidence of neurologic injury is still much higher than with
idiopathic deformity. Standard treatment measures for neurologic loss
following surgery for congenital scoliosis

include
maximizing spinal cord perfusion and oxygenation, considering
diminishing correction or removing instrumentation, and evaluation for
direct mechanical compression of the cord by appropriate imaging such
as CT myelogram.

Treatment of congenital scoliosis may lead to a
worsening of deficiencies in existing trunk height. Generally the loss
of trunk height is well tolerated (103,166,167)
and is not considerable compared with the deformity that would have
resulted if the condition had been allowed to progress. Goldberg et al.
have brought attention to the shortened spine height associated with
congenital spinal deformity and diminished trunk and spinal dimensions
after spinal arthrodesis (72,107). Campbell et al. have described thoracic insufficiency following spinal arthrodesis for congenital scoliosis (104)
associated with fused ribs. Patients with a preexisting short trunk,
chest wall anomalies, or other causes of restrictive lung disease may
not tolerate the further shortening associated with an extensive spinal
arthrodesis. In this select group, procedures such as instrumentation
without fusion or expansion thoracostomy with VEPTR insertion may be
preferable to an early extensive arthrodesis, acknowledging that
eventual arthrodesis will be needed at a later stage of growth.
Diminished height of the spine, thoracic insufficiency syndrome, and
the effects of early fusion (168) may not become apparent until the adolescent period, when the normal thoracic volume increases greatly.

Degeneration of adjacent segments and difficulties with
the transition zone between spinal fusion and the mobile unfused spine
are acknowledged and documented phenomena associated with spinal
fusions in all ages (169,170,171).
A similar phenomenon of increased instability and degeneration of
adjacent segments occurs adjacent to congenital fusions that extend
into the cervical spine in the Klippel-Feil association (27,29,106,172).
Adjacent segment degeneration or transition zone instability can be
seen after fusion for congenital scoliosis; they may manifest as pain
or neurologic deficit associated with spinal cord compression at the
degenerated level caused by hypertrophy of the hypermobile disc,
ligamentum flavum, or associated osteophytes. A consideration of the
possible occurrence of adjacent segment degeneration should be factored
into the choice of surgical options in congenital scoliosis.