The routine use of magnetic resonance imaging in the evaluation
of a patient with scoliosis remains controversial. While intraspinal
pathology is a well-known causative factor in the development of
scoliosis2-5, it is usually associated
with an atypical left-sided thoracic curve6,
a young age at presentation1,
or an abnormal neurologic examination7.
Using magnetic resonance imaging, O’Brien et al.8 found no evidence of abnormalities
in a series of thirty adolescent patients with a typical right-sided thoracic
curve and a normal neurologic examination. Samuelsson et al.9, however, demonstrated an 8% prevalence
of spinal syringomyelia in twenty-six patients with an average age
of twelve years. Therefore, they advocated the routine use of magnetic
resonance imaging in the work-up of a patient with idiopathic scoliosis.
In an attempt to settle this controversy, we analyzed magnetic
resonance images made as part of a preoperative work-up in a consecutive
series of 327 patients who presented with adolescent idiopathic scoliosis.
The purpose of the study was to determine the prevalence of intraspinal
anomalies and the clinical relevance of routine magnetic resonance imaging
in the evaluation and treatment of patients with adolescent idiopathic
scoliosis and a normal neurologic examination.
Beginning in December 1991, patients between the ages of ten
and twenty years with idiopathic scoliosis requiring operative treatment
were enrolled in the study. The other criteria for inclusion in
the study consisted of a negative clinical history (no pain, numbness,
or weakness in the back or lower extremities), a normal physical
examination, and a typical adolescent idiopathic scoliosis curve
pattern. From December 1991 to March 1999, 327 consecutive patients
fulfilled all requirements for participation in the study.
Two additional patients who were seen during this study period
had a curve pattern that was similar to that of adolescent idiopathic
scoliosis, but one presented with intermittent back pain and the
other had vocal cord paralysis at the examination prior to the operation.
They had a normal motor and sensory examination but had asymmetrical
superficial abdominal reflexes. Both had a major holocord syrinx
requiring neurosurgical decompression. Because of their symptoms
as well as the abnormal abdominal reflexes, they were excluded from
the series.
Prior to spinal arthrodesis, all patients underwent a comprehensive
evaluation, which consisted of a history, review of systems, and
physical examination by a pediatric resident and a staff pediatrician. If
any abnormal findings were detected, the patient was further evaluated
with other diagnostic tests as indicated.
The orthopaedic examination consisted of a complete evaluation
of the back, including plumb-line assessment for decompensation
of the trunk and identification of apical trunk rotation and rib
prominence. Motor strength and sensation in the upper and lower
extremities as well as deep tendon and abdominal reflexes were assessed.
The examination was performed first by an orthopaedic resident or fellow
and then by a staff orthopaedic surgeon. Any patients with a clinical
history that aroused suspicion of a neurologic abnormality, such
as a history of radicular pain or physical findings of muscle atrophy,
sensory or motor deficits, or reflex changes, were referred to a
pediatric neurologist for further evaluation.
Magnetic resonance imaging of the spine was performed with use
of a Signa 1.5-tesla unit coil (General Electric Medical Systems,
Milwaukee, Wisconsin) with a dedicated posterior phased array cervicothoracolumbar
spine coil (General Electric Medical Systems). The parameters for
the sequences varied somewhat on the basis of the patient’s
size and heart rate and the presence of braces. The usual study
included sagittal images, made in two segments (a 30-cm field of
view for each) with use of T1-weighted images (repetition time,
450 msec; echo time, 15 msec; echo train length, four) and fast-spin-echo
T2-weighted images (repetition time, 4000 msec; effective-echo time,
80 msec; echo train length, twelve to sixteen) from the skull base
to the midthoracic region, which were cardiac gated, and from the
midthoracic region to the sacrum. Slice thickness ranged from 3.5
skip 0 to four skip 1 mm. Axial images were obtained with use of
generous (8 to 10-mm) gradient-echo sections (repetition time, 400
msec; echo time, 20 msec; flip angle, 30°) from the skull base to
the midthoracic region. These were cardiac gated and had to be changed
to fast-spin-echo T2-weighted images for children with extensive
orthodontics because of the metallic artifact caused by the braces.
Fast-spin-echo T2-weighted images were acquired from the midthoracic
region to the sacrum.
A fast-spin-echo T2-weighted sequence was performed in the midthoracic
region with use of a large (30-cm) field of view at 3-mm sections.
The matrix size was 256 ¥ 192, except on the coronal images, where
the matrix ranged from 256 ¥ 192 to 512 ¥ 256.
The appearance of the cord, the level of the conus medullaris, and
the presence or absence of syringomyelia, diastematomyelia, or Chiari
malformations were evaluated on the magnetic resonance images by
the radiologists. If an abnormality was suspected, either additional
magnetic resonance imaging studies were performed as needed or myelography
was performed, followed by computed tomography, to further evaluate
the abnormality.
The study included 327 consecutive patients with adolescent idiopathic
scoliosis. There were eighty-two males and 245 females. The average
age of the patients was thirteen years and six months (range, ten
to nineteen years). There were 171 right thoracic curves, seventy-two
thoracolumbar curves, and eighty-four double major curves. The average Cobb
angle of the major curve measured 57° (range, 40° to 98°). Sensation,
motor strength, and reflexes were normal in all of the patients.
Seven patients, all with a right thoracic curve, had an abnormality
noted on magnetic resonance images. These abnormalities included
a thoracic syrinx in two patients (0.6%) and an Arnold-Chiari type-I
malformation in four (1.2%). One patient had a fatty deposit
in the tenth thoracic vertebral body. All patients with a radiographic
abnormality were evaluated by a pediatric neurologist and a neurosurgeon.
The fatty vertebral body was not considered important and did not
require additional work-up.
The consulting neurosurgical team did not think any of the other
six abnormalities needed decompression because none of the patients
had pain, radiculitis, radiculopathy, or an Arnold-Chiari malformation
with tightness at the foramen magnum. Tightness at the foramen magnum
was defined as protrusion of the cerebellar tonsils into the foramen
such that there was essentially no free room. This could lead to
obstruction of the normal flow of cerebrospinal fluid, thereby increasing
pressure in the ventricular system and possibly enlarging a preexisting
cyst or syrinx.
In four cases, the interpretation of the initial magnetic resonance
images was equivocal, necessitating additional radiographic imaging.
In one, the interpretation was a possible dural sac division in the
cervical spine. The cervical magnetic resonance imaging study was
repeated for this patient, and it was found to be normal. In another
patient, areas of diminished signal intensity were noted in the
spinal canal. A computed tomographic myelogram was made, and the
results were found to be normal. In the remaining two patients,
a dural sac abnormality was considered to be possible on the basis
of areas of diminished signal intensity on a limited study. Additional
studies were recommended; however, the patients and their families
refused to undergo an additional work-up. During a second review,
the differences in signal intensity were considered to be a problem
of signal averaging and were not thought to be important.
Three hundred and twenty-four patients in the study proceeded
with the planned spinal fusion and segmental instrumentation. Three
patients and their families decided against the operation. The operation
consisted of posterior spinal fusion in 282 patients, anterior spinal
fusion in fourteen, and combined anterior-posterior spinal fusion
in twenty-eight. The choice of either isolated anterior spinal fusion
or isolated posterior spinal fusion depended on the type and magnitude
of the curve as well as on the surgeon’s preference. Combined anterior
and posterior spinal fusions were performed on large rigid curves
to obtain better correction.
The average preoperative curve measured 57° (range, 40° to 98°).
The average postoperative curve measured 14° (range, 0° to 49°),
with an average curve correction of 76%. Intraoperative somatosensory
evoked potential monitoring was performed on all patients from the
beginning of the operation to its conclusion, with no changes noted. Motor
evoked potentials were not routinely monitored. The Stagnara wake-up
test was performed after the instrumentation was applied in all
patients; there was no detectable neurologic deficit. There were
no complications in the immediate perioperative period.
In 1996, Shen et al.10 reported
on a series of seventy-two consecutive patients with adolescent
idiopathic scoliosis and a completely normal neurologic exam. All
patients underwent routine preoperative magnetic resonance imaging
of the spine. An abnormality was noted in three patients (4%).
Two patients had an Arnold-Chiari type-I malformation, and one had
a fatty deposit in the tenth thoracic vertebral body. The abnormalities
did not require intervention and did not alter the management of
the scoliosis. Winter et al.11 evaluated
140 patients and reported that four (3%) had a small syrinx.
None of these patients required neurosurgical intervention prior
to their spinal surgery.
We included some of the data from the study by Shen et al.10 and increased our sample size to
327 patients; we believe that our series represents the largest
number of patients evaluated prospectively with magnetic resonance
imaging prior to arthrodesis for the treatment of adolescent idiopathic
scoliosis. All 327 patients had a normal neurologic examination, including
normal superficial abdominal reflexes. In these 327 patients, the
prevalence of positive findings on the magnetic resonance images
was 2% (seven patients). Two patients had a small thoracic syrinx,
four had a mild Arnold-Chiari type-I malformation, and one had a
fatty deposit within the tenth thoracic vertebral body. None of
these patients required neurosurgical intervention or an additional
work-up prior to undergoing spinal arthrodesis. All patients
who underwent spinal fusion had an uneventful course with normal
intraoperative somatosensory evoked potentials and a normal wake-up
test. None of those patients showed any delayed neurologic sequelae
at the time of follow-up.
The results of our study underscore the importance of a thorough
history and physical evaluation, including a detailed neurologic
examination, for patients with adolescent idiopathic scoliosis.
The findings can be as subtle as an asymmetric superficial abdominal
reflex even though the sensitivity and specificity of this reflex
are of questionable accuracy and reproducibility11-13.
On the basis of this study, we believe that preoperative magnetic
resonance imaging is neither necessary nor indicated for an otherwise
healthy and neurologically intact patient with adolescent idiopathic
scoliosis.
Note: The authors thank Tanaya Parikh for her assistance in the
data collection and Douglas Mintz, MD, a radiologist, for his magnetic
resonance imaging analysis of patients in this study.