Thoracic insufficiency syndrome is the inability of the thorax to support
normal respiration or lung
growth1. The
syndrome may be present when there are extensive fused ribs associated with
congenital thoracic scoliosis that adversely affects thoracic volume,
function, and growth with an equally adverse effect on the function and growth
of the lungs (Figs. 1-A and
1-B). Progressive thoracic insufficiency syndrome due to
three-dimensional thoracic deformity and dysfunction can be characterized on
the basis of the respiratory history and the findings on physical examination,
radiographs, computed tomography, and pulmonary function studies. Ideally,
treatment should improve thoracic volume and function and maintain these gains
during growth1.
Traditional spine surgery with instrumentation achieves some correction of
rigid congenital curves, but the mechanical advantage is too poor to
effectively expand the lateral part of the rib cage constricted by rib fusion.
Thus, the global three-dimensional thoracic deformity remains unaddressed.
Since 1987, we have evaluated more than 500 patients with thoracic
insufficiency syndrome due to severe malformations of the thorax. The
anomalies in these patients included extensive congenital scoliosis with
either fused or absent ribs, Jarcho-Levin syndrome, Jeune asphyxiating
thoracic dystrophy, and progressive infantile scoliosis. As an alternative to
spine surgery, we developed several types of expansion thoracoplasties and
have treated more than 150 of these children with those procedures. For
patients with severe progressive congenital thoracic scoliosis with fused
ribs, we developed an expansion thoracoplasty, termed an opening wedge
thoracostomy, that directly treats the segmental hypoplasia of the
hemithorax due to fused ribs while indirectly correcting the scoliosis. The
thoracostomy correction is stabilized with a chest-wall device known as a
vertical, expandable prosthetic titanium rib. Serial expansion of the
device through limited incisions maintains the correction during growth. For
our patients, the goals of treatment were to restore lost thoracic volume by
straightening the so-called angulated thorax through acute surgical
lengthening of the constricted, concave hemithorax, thereby equilibrating the
thorax with indirect correction of the scoliosis in order to restore symmetry
in the coronal, sagittal, and transverse planes. It is also necessary to
maintain the initial correction during growth. Periodic prosthetic lengthening
preserves thoracic spine growth so that the thorax can gain additional volume
through increases in spinal height. The purpose of this study was to assess
these patients and determine their response to treatment.
Since 1990, forty-one patients with fused ribs and congenital scoliosis
were treated with opening wedge thoracostomy, and twenty-seven of them were
followed for a minimum of two years (see Appendix). No patient was lost to
follow-up. Patients were enrolled for treatment prospectively after a
tri-specialty evaluation by a pediatric orthopaedic surgeon, a pediatric
general surgeon, and a pediatric pulmonologist. The criteria for acceptance
for treatment with opening wedge thoracostomy included an age of six months to
skeletal maturity, progressive thoracic insufficiency syndrome, more than a
10% reduction in the height of the hemithorax on the concave side of the curve
compared with the height of the contralateral hemithorax, and three or more
anomalous vertebrae with three or more fused ribs at the apex of the
deformity. The surgery was recommended only when there was unanimous agreement
among all three specialists. The study was approved by our institutional
review boards, and all parents gave informed consent for the treatment of
their child.
Preoperative Evaluation
Clinical respiratory insufficiency in these children was identified on the
basis of an ease of fatigability, an inability to keep up with peers, or
numerous episodes of pneumonia or bronchitis. Coexisting lung disease, such as
asthma, was also documented. Preoperative vital signs, including pulse,
respiratory rate, blood pressure, and oxygen saturation as measured with a
pulse oximeter, were determined. Shoulder depression and the range of motion
of the shoulders were evaluated with the patients sitting. Sitting and
standing height were measured as well. Primary breathing by the diaphragm was
assessed by measuring the abdominal circumference at both inspiration and
expiration, and secondary
breathing1 was
evaluated on the basis of the respiratory changes in the circumference of the
chest at the nipple line. Deformities of the thorax were recorded, and the
extent of any flail segments with paradoxical motions was measured.
Thoracic deformity in the coronal and sagittal planes was evaluated on
radiographs, and thoracic deformity in the transverse plane was evaluated with
computed tomography. Standing or sitting anteroposterior and lateral
radiographs of the entire spine, including the chest; a cross-table lateral
radiograph of the chest and spine; and supine lateral bending radiographs of
the spine to assess curve flexibility were made. Supine spiral tomograms of
the spine were also made. Supine anteroposterior and lateral radiographs were
made with a scanogram ruler adjacent to the patient. The tube-to-table
distance was 102 cm, and the table-to-film distance was an additional 7.6 cm.
Unenhanced computed tomography of the thorax was performed at 1-cm intervals
until 1994 and at 0.5-cm intervals thereafter. Magnetic resonance imaging of
the cervical, thoracic, and lumbar spines was performed for all patients.
Those with a tethered spinal cord were treated surgically for this problem
prior to our treatment.
Usually, only patients who were six years of age or older were able to
cooperate with pulmonary function testing with spirometry, so preoperative
testing was limited. All percent-of-normal values derived with pulmonary
function testing were based on arm-span measurements. Capillary blood gases
were analyzed in all patients. Electrocardiograms were also made for all
patients, and echocardiograms were done when clinically indicated.
After the initial implantation, all examinations and tests were repeated
with the exception of supine bending radiographs, magnetic resonance imaging,
and echocardiograms.
Before the device-expansion operations, standing or sitting anteroposterior
and lateral radiographs of the entire spine, including the chest; a
cross-table lateral radiograph of the chest and spine; and supine lateral
bending radiographs of the spine and computed tomography scans were made. All
except the lateral bending radiographs and computed tomography scans were
repeated after those operations.
Radiographic Measurements
Congenital scoliosis, thoracic kyphosis, and lumbar lordosis were measured
with the Cobb
method2. All
radiographic measurements were made by one of us. Head decompensation was
assessed on weight-bearing radiographs by measuring the lateral displacement,
in centimeters, from the center sacral line to the posterior spinous process
of C73. Trunk
decompensation was assessed by measuring the lateral displacement, in
centimeters, of the center of the transverse diameter of the thorax at the
level of T6 from the central sacral
line4. A central
rigid curve segment, defined as the portion of the Cobb angle curve that did
not change in magnitude on preoperative supine lateral bending radiographs,
was measured.
To assess the lateral deviation of the spine in congenital scoliosis, we
developed an interpedicular line ratio method
(Fig. 2). The measurement error
of this technique was assessed with a modification of the approach of
Facanha-Filho et
al.5. Fifty
randomized radiographs of extensive congenital scoliosis with fused ribs in
which the end vertebrae of the curve had been identified were measured by
three experienced spine surgeons, who then repeated the measurements three
weeks later. The same lateral pedicle landmark was identified by all three
observers on forty of the fifty curves, and the mean intraobserver variance of
the interpedicular ratio of the curves was 0.11 (range, 0.09 to 0.13) with a
95% confidence level of ±0.03. This meant that a change of 0.17 between
measurements of this ratio by the same observer could be due to measurement
error; thus, we considered a change of 0.2 in the ratio as a true change. The
mean interobserver variance error was 0.14.
The type of congenital scoliosis was classified according to the criteria
of MacEwen6, as
modified by Winter et
al.7, for each
individual level, with identification of unilateral unsegmented bars and
hemivertebrae and the location and extent of rib fusion. In order to define
the extent of the congenital deformity of the spine, a percentage was
calculated by dividing the number of thoracic vertebrae with congenital
anomalies by the total number of thoracic vertebrae.
The height of each hemithorax was measured on weight-bearing radiographs,
and the space available for the
lung1 was calculated
on the basis of distance from the proximal rib to the dome of the
hemidiaphragm. To better appreciate changes in the constricted hemithorax, we
found it helpful to rotate standard anteroposterior radiographs 90° on the
view box with the concave side of the curve always upward to show what we
termed a false lateral decubitus
view1.
Distances on radiographs were measured with a ruler and then corrected by
comparison with the scanogram image on the radiograph. Additional accuracy was
obtained by determining residual image magnification for each type of
measurement by making multiple test radiographs of a small female cadaver
thorax with use of the just described technique, comparing radiographic and
physical measurements of known landmarks. The image magnification for
measurements of hemithorax height was 6%. The transverse diameter of the
thorax at T6 at the inner border of the ribs was measured on radiographs with
a ruler, with adjustment according to the scanogram scale on the radiograph
and additional correction by an 8% image magnification factor. The sagittal
diameter of the thorax on the lateral radiograph was also assessed in this
manner, with measurement from the posterior border of the sternum to the
anterior surface of the posterior ribs at the T6 level, with a 12% image
magnification correction factor. The thoracic spinal height was determined as
the distance between the midpoints of the superior end plate of T1 and the
inferior end plate of T12, with an image magnification correction factor of
8%. The spinal length was determined for all curves by identifying the central
axis between pedicles for each vertebra and measuring along them with a
flexible ruler that had a self-contained centimeter scale (Truflex Dual
Graduated Flexible Curve; Alvin and Company, Rancho Cordova, California); the
image magnification correction factor was 8%.
Preoperative, postoperative, and follow-up radiographs were all made at our
institution and were standardized as described above. Anteroposterior
radiographs from other institutions, used to determine progression by
documenting the early values for the scoliosis and the space available for the
lung, were not standardized, but the measurements that were used were not
subject to magnification error. The anteroposterior radiographic measurements
defined the thoracic deformity in the coronal plane, and the lateral
radiographic measurements defined the sagittal plane deformity.
Computed tomography scans were analyzed to evaluate the changes with time
in the shape of the thorax and were used to define the transverse plane
deformity. The slice closest to the apical vertebra of the curve, showing the
most distortion of the thorax, was chosen on each preoperative computed
tomography scan, and the equivalent slice was identified on the follow-up
scan. The posterior hemithoracic symmetry ratio, thoracic rotation, and spinal
rotation1 were
measured preoperatively and at the time of follow-up. Computed tomography
scans made prior to the surgery were also available for eighteen patients, and
these were assessed for preoperative progression of transverse-plane thoracic
deformity in the same fashion.
The paired Student t test was used to evaluate differences between
preoperative and postoperative values. Differences in the results between
groups differentiated according to age or whether a previous spine fusion had
been performed were assessed with analysis of variance. For both tests, a p
value of <0.05 indicated a significant difference.
Surgical Technique
The vertical, expandable prosthetic titanium rib has three components: a
superior rib cradle to attach to the cephalad osseous rib or ribs, a hollow
central component called the rib sleeve, and an inferior rib cradle
to allow later expandability. For curves extending into the lumbar spine, we
use a hybrid device in which the inferior rib cradle is replaced with a
modified spinal rod that is inserted into the rib sleeve
(Fig. 3). The first, prototype
devices were manufactured by Techmedica, Camarillo, California, from 1989 to
1994. The prosthesis now in use is a fourth-generation device made by Synthes
Spine, West Chester, Pennsylvania (United States Patent numbers 5,092,889 and
5,261,908, European Patent number 0530177, and others) and is available under
the Food and Drug Administration Humanitarian Device Exemption.
A modified thoracotomy incision is made, extending distally gently in a J
shape. An opening wedge thoracostomy is then performed at the apex of the
thoracic constriction, and the interval is gradually widened (Figs.
4-A, 4-B,
4-C, and
4-D) to lengthen the
constricted hemithorax. The acute correction is stabilized by the prosthesis
(Fig. 4-C). A hybrid rib
prosthesis (Fig. 3) is used for
thoracolumbar scoliosis. That device is attached to cephalad ribs, with the
lower portion attached to the spine or the iliac crest. At scheduled intervals
of four to six months following the initial implantation, the devices are
expanded through limited incisions in an outpatient surgery setting to
accommodate growth (Fig.
4-D).
Details of the surgical technique have been published
previously8.
Clinical Results
The twenty-seven patients were followed for a mean of 5.7 years (range, two
to twelve years). The mean preoperative curve progression, calculated by
comparing the earliest referral radiographs available prior to the surgery
with the immediate preoperative radiographs, was 18° (range, 5° to
60°). The mean yearly rate of progression before the operation was
15°/yr (range, 2° to 50°/yr). The mean age of the patients at the
time of the initial surgery was 3.2 years (range, 0.6 to 12.5 years).
Seventeen patients were male, and ten were female. Four patients had a
myelomeningocele. Twenty-five patients had a unilateral unsegmented bar with
multiple convex hemivertebrae, and two had congenital wedged vertebrae. All
had fused ribs at the concavity of the curve. Three patients, including one
with a myelomeningocele, had a tethered spinal cord. One patient had
respiratory insufficiency and required support with oxygen by continuous
positive airway pressure. The remaining patients had normal clinical
respiratory function.
The mean duration of the initial surgery was 4.1 hours (range, 1.7 to 6.5
hours). The mean operative time for device-replacement procedures was 2.5
hours (range, 0.4 to 8.8 hours). Expansion procedures required a mean of
sixty-five minutes (range, twenty-two to 145 minutes). The mean estimated
blood loss during the initial surgery was 57 mL (range, 4 to 220 mL), or 7% of
the total blood volume. Thirteen patients required a blood transfusion during
the first two postoperative days. The mean estimated blood loss was 49 mL
(range, 5 to 1578 mL) during the device-replacement procedures and 7 mL
(range, 1 to 30 mL) during the expansion procedures. The mean duration of
hospitalization for the initial procedure was fourteen days (range, nine to
thirty-five days), with a mean of 7.3 days (range, two to twenty-eight days)
in the pediatric intensive care unit. A total of thirty-three devices were
implanted, with six patients requiring implantation of devices on both sides
in a staged fashion. Seventy-one device-replacement procedures and 176
expansion procedures were performed. The patients had a mean of 10.4
procedures (range, five to twenty procedures) in total and 1.8 procedures per
year.
All patients had unequal shoulder levels both preoperatively (mean
inequality, 0.48 cm; range, 0 to 4 cm) and post-operatively (mean inequality,
1.15 cm; range, 0 to 3 cm). Shoulder abduction on the operatively treated side
did not change following the surgery. Preoperatively, seven patients had trunk
alignment within 1 cm of the midline, whereas at the time of final follow-up
twelve patients had trunk alignment within 1 cm of the midline. As measured
according to the lateral deviation from the center sacral line, head
decompensation was a mean of 2 cm (range, 0 to 6.7 cm) preoperatively and 2.6
cm (range, 0 to 6.3 cm) at the time of final follow-up. The mean sitting
height was 48.9 cm (range, 36.5 to 68.5 cm) before the surgery and 60.1 cm
(range, 50 to 73 cm) at the time of follow-up. The mean standing height was
86.5 cm (range, 59.5 to 142 cm) preoperatively and 117.5 cm (range, 86 to 149
cm) at the time of follow-up.
Pulmonary Results
There was no significant change in primary or secondary respiration as
measured by the change in chest or abdominal circumference with breathing.
Oxygen saturation levels also showed no significant changes. The respiratory
rates did change, and they were compared with normal rates for individuals of
the same age. The normal respiratory rate is thirty to eighty breaths per
minute at birth, twenty to forty breaths per minute up to five years of age,
fifteen to twenty-five breaths per minute from the age of six to twelve years,
and fifteen to twenty breaths per minute after fifteen years of age and in
adulthood9. The mean
preoperative respiratory rate of our patients was twenty-eight breaths per
minute (range, twenty to sixty-four breaths per minute), with three patients
having a respiratory rate that was greater than normal for their age. All
three patients had a normal respiratory rate at the time of follow-up. There
was a mean decrease in the respiratory rate to twenty-two breaths per minute
(range, twenty to thirty-six breaths per minute), although a
greater-than-normal respiratory rate developed in one patient. One patient was
weaned off continuous positive-airway-pressure oxygen support, had his
tracheostomy tube removed, and went on to normal activities while breathing
room air. One patient, who had required only room air preoperatively, needed
nighttime respiratory support with continuous positive airway pressure at the
time of follow-up. The need for nighttime nasal oxygen developed in another
patient. The remaining twenty-four patients remained stable with regard to
their ability to function on room air, were active, had minimal respiratory
illnesses, and kept up with their peers in play activities.
Only three patients were old enough to cooperate with pulmonary function
testing with spirometry preoperatively. During the follow-up period, an
additional nineteen became mature enough to have standard pulmonary testing.
These twenty-two patients had a mean forced vital capacity at the time of the
last follow-up, at a mean of 5.9 years (range, two to twelve years), of 0.89 L
(range, 0.36 to 2.07 L) and 49% of the predicted normal value (range, 25% to
95% of the predicted normal value). These twenty-two patients were separated
into three groups. Group A consisted of eight patients who were less than two
years of age at the time of their initial surgery, which is the age when there
is the most rapid lung growth by lung alveolar cell
multiplication10,11;
Group B consisted of eleven patients who were two years of age or older at the
time of the initial surgery and had no history of spine fusion; and Group C
consisted of three patients who were two years of age or older at the time of
the initial surgery and had had a previous spine fusion. The mean age at the
time of follow-up was 8.5 years (range, 4.7 to 12.6 years) in Group A, 9.4
years (range, five to 14.8 years) in Group B, and twelve years (range, 9.2 to
15.9 years) in Group C. Although there was a trend for the mean vital
capacities in milliliters at the time of follow-up to be higher for the
patients who were two years of age or older at the time of the surgery, with
the numbers available there were no significant differences between groups (p
= 0.74) (Fig. 5-A). The
percentage of the predicted normal vital capacity, however, was significantly
higher for Group A than it was for Group B (p < 0.001). There was also a
difference in the predicted normal vital capacity between Group B and the
three patients in Group C, but this was not significant with the numbers
available (p = 0.34) (Fig.
5-B). These findings suggest that patients treated during the
period of most rapid lung growth had the most favorable vital capacity for
their size at the time of follow-up.
Interval pulmonary function studies were analyzed to determine trends with
regard to changes in vital capacity in the period following treatment. Sixteen
of the twenty-two patients with pulmonary function tests had such interval
studies, at a mean of 3.1 years (range, two to 6.7 years) postoperatively. The
first postoperative test demonstrated a mean vital capacity of 0.679 L (range,
0.37 to 1.7 L), or 49% (range, 33% to 68%) of the predicted normal vital
capacity, whereas the mean vital capacity at the time of follow-up was 0.91 L
(range, 0.51 to 2.1 L), or 47% (range, 25% to 66%) of the predicted normal
vital capacity. Seven patients in Group A with interval studies had a mean
vital capacity of 0.51 L (range, 0.37 to 0.81 L) demonstrated by the first
postoperative test, with an increase to 0.72 L (range, 0.51 to 1.04 L) at the
time of follow-up, at a mean of 2.75 years (range, two to 4.58 years). The
percentage of the predicted normal vital capacity remained stable, with a
change from 50.7% to 53%. Seven patients in Group B with interval studies had
a mean vital capacity of 0.69 L (range, 0.48 to 1.16 L) at the first test,
with an increase to 0.98 L (range, 0.58 to 1.54 L) at the time of follow-up,
at a mean of 3.1 years (range, 0.5 to 6.67 years). The mean percentage of the
predicted normal vital capacity also remained stable, changing from 44% to
45%. In these two groups of patients, all changes in the volume of vital
capacity were significant (p < 0.001), but the changes in the percentage of
the predicted normal vital capacity were not (p > 0.69). Two patients in
Group C had both preoperative and follow-up studies; the mean vital capacity
at the first test was 1.036 L (0.39 and 1.68 L), with a mean increase to 1.335
L (0.59 and 2.08 L) at the time of follow-up. However, both had a decrease in
the percentage of the predicted normal vital capacity, from a mean of 48% (36%
and 60%) to 38% (29% and 47%) at the time of follow-up, at a mean of 3.7 years
(3.3 and four years).
Analysis of capillary blood gases showed a mean Po2
level of 66.5 mm Hg (range, 37.7 to 131 mm Hg) preoperatively and 74.7 mm Hg
(range, 56 to 98.8 mm Hg) at the time of follow-up. The mean
Pco2 levels were 39.9 mm Hg (range, 31.9 to 66 mm Hg)
preoperatively and 38 mm Hg (range, 32.4 to 48.6 mm Hg) postoperatively. These
changes were not significant (p = 0.15). The normal capillary
Po2 level is usually 10 mm Hg lower than the arterial
Po2 level when the arterial Po2
level is <70 mm Hg; when the arterial Po2 level is
>70 mm Hg, the capillary Po2 level is
unreliable12. The
capillary Pco2 level correlates well with the arterial
Pco2 level (normal level, 35 to 45 mm
Hg)12.
Three-Dimensional Thoracic Deformity
Coronal-Plane Thoracic Deformity and Growth (see Appendix)
In twenty-five of the twenty-seven patients, a mean of 4.7 hemivertebrae
(range, one to twelve hemivertebrae) were present on the convex side. These
patients had a unilateral unsegmented bar on the concave side of the curve in
the thorax, with involvement of a mean of 4.2 vertebrae (range, two to twelve
vertebrae). The other two patients had wedged vertebrae with close
approximation of the pedicles, but no bars were seen with tomography.
Additional congenital anomalies of the thoracic spine in our patients included
wedged or butterfly vertebrae cephalad and caudad to the unilateral
unsegmented bars. The mean percentage of the total number of thoracic
vertebrae with various congenital anomalies was 66% (range, 33% to 100%), and
the patients had a mean of 7.3 fused ribs (range, three to twelve fused ribs).
It was extremely difficult to determine with certainty, on examination of the
plain radiographs, whether rib refusion had occurred after the surgery, but
gaps between the osteotomized ribs stayed constant in all patients, as seen on
the follow-up radiographs.
The mean number of vertebrae included in the Cobb angle was 10.6 (range,
six to fifteen). Preoperatively, the mean Cobb angle on the weight-bearing
anteroposterior radiographs was 74° (range, 35° to 140°). The
curves decreased to a mean of 56° (range, 20° to 114°) immediately
postoperatively and, after multiple lengthenings of the device, further
decreased to a mean of 49° (range, 4° to 84°). Both the
preoperative curve progression and the postoperative improvement were
significant (p < 0.0001). The mean number of vertebrae in the rigid portion
of the thoracic curve was 6.9 (range, three to eleven), and the mean
percentage of vertebrae in the Cobb angle that consisted of the rigid portion
was 65%. This rigid portion of the curve originally was thought to have little
potential for correction but did go on to show improvement. The rigid portion
of the curve was a mean of 56° (range, 5° to 130°) preoperatively,
and this decreased to a mean of 42° (range, 0° to 79°) at the time
of follow-up (Figs. 6-A and
6-B). The initial, preoperative, and follow-up values for both the
mean interpedicular line ratio and the space available for the lung were found
to differ significantly (p < 0.05) (Fig.
7). The height of the convex hemithorax was found to have
decreased by only a mean of 2 mm immediately postoperatively, suggesting that
the increase in the space available for the lung ratio was primarily due to an
increase in the height of the concave hemithorax and not to a decrease in the
height of the convex hemithorax. The change in the Cobb angle, interpedicular
line ratio, and space available for the lung ratio between the preoperative
and follow-up assessments was analyzed for Groups A, B, and C
(Table I), and all changes were
significant (p < 0.05) except for that in the space available for the lung
ratio in Group C (p = 0.09). Comparison of the changes in the Cobb angle,
interpedicular line ratio, and space available for the lung ratio between the
groups showed no significant differences, with the numbers available (p >
0.05).
The mean height of the thoracic spine was 10.9 cm (range, 6.7 to 14.9 cm)
preoperatively, and it increased to 11.7 cm (range, 7.5 to 19.8 cm)
immediately after the initial implantation of the device as a result of a
combination of both curve correction and distraction of the spine. At the time
of follow-up, it had increased to 15 cm (range, 11 to 22 cm) (p < 0.0001).
The thoracic spinal height increased a mean of 0.71 cm/yr (range, 0.03 to 1.46
cm/yr). The mean preoperative length of the thoracic spine was 11.7
cm (range, 7.4 to 20.2 cm), and it increased to only a mean of 12.3 cm (range,
8.5 to 20.2 cm) immediately following the surgery (p = 0.22); however, it had
increased significantly to 15.7 cm (range, 9.6 to 23.2 cm) at the time of the
last follow-up (p < 0.005). Over the follow-up period, the yearly mean
change in length, which was probably affected less by curve correction and
more by growth, was 0.7 cm/yr (range, 0.2 to 1.37 cm/yr), or 6%/yr (range,
5.8% to 7.8%/yr) of the preoperative length. According to Dimeglio and
Bonnel13, the rate
of growth of the normal thoracic spine is 1.4 cm/yr from birth to the age of
five years, 0.6 cm/yr from five to ten years of age, and 1.2 cm/yr from ten to
sixteen years of age. The mean height of the normal thoracic spine is 12 cm at
birth, 18 cm at five years of age, 21.5 cm at ten years of age, and 28 cm for
males and 26.5 cm for females at maturity. The growth rate of a thoracic spine
with untreated congenital scoliosis is unknown. The mean width of the thorax
was 12.7 cm (range, 9.2 to 19.3 cm) preoperatively, and it had increased to
15.6 cm (range, 11 to 22.4 cm) at the time of follow-up (p < 0.0001). The
mean yearly increase in the thoracic width was 0.51 cm/yr (range, 0 to 1.77
cm/yr).
Sagittal-Plane Thoracic Deformity and Growth
The mean thoracic kyphosis was 14° (range, -50° to 61°)
preoperatively, and it had increased to a mean of 33° (range, -15° to
85°) at the time of the last follow-up. The mean lumbar lordosis was
20° (range, -40° to 50°) preoperatively, and it had increased to a
mean of 26° (range, -32° to 63°) at the time of follow-up.
Preoperatively the sagittal depth of the thorax was a mean of 10 cm (range,
5.8 to 15 cm), and at the time of the last follow-up it was a mean of 13.6 cm
(range, 9.1 to 18.3 cm) (p < 0.0001). The mean yearly growth of the
sagittal diameter was 0.63 cm/yr (range, 0.02 to 2.04 cm/yr).
Transverse-Plane Thoracic Deformity
Computed tomography was performed for twenty-five patients immediately
before the surgery and at the time of the last follow-up. Eighteen of those
patients also had computed tomography performed at a mean of 10.9 months
(range, two to twenty-three months) before their first surgical procedure.
There was a trend toward an increase in the mean posterior hemithoracic
symmetry ratio between the initial and immediate preoperative computed
tomography assessments, suggesting progression of the deformity. The mean
posterior hemithoracic symmetry ratio decreased after treatment, but the
change was not significant (p > 0.23). The spinal and thoracic rotation
angles did not change significantly either before or after treatment (p >
0.11). Differences between age groups also were not significant (p > 0.06).
These findings suggest that the transverse-plane thoracic deformity was stable
both in the period before the surgery and during treatment.
Secondary Thoracic Insufficiency Syndrome
Two of the four patients with myelomeningocele had preoperative lumbar
kyphosis. This kyphosis is a risk factor for what we call secondary
thoracic insufficiency, and one patient had clinical signs of this
disorder (Figs. 8-A and 8-B).
With secondary thoracic insufficiency, diaphragmatic function is compromised
because of collapse of the base of the chest onto the pelvis. The diaphragm
must contract against increased abdominal pressure, in essence doing a
"push-up" against body weight, with the risk of respiratory
fatigue. Clinically, these patients have what we term a marionette
sign, with synchronous bobbing of the head with respiration. The
diagnosis can be confirmed with fluoroscopy of the diaphragm performed while
the patient is sitting. Conditions that raise abdominal pressure in these
patients, such as constipation, can cause respiratory distress. Successful
treatment increases the distance between the pelvis and the thorax to allow
increased diaphragmatic excursion. The other patient with lumbar kyphosis, who
was treated only with a thoracic device that did not address the lumbar
deformity, had progression of the lumbar kyphosis from 20° to 32°. She
did not have the marionette sign, but she had become dependent on nighttime
continuous positive airway pressure at the time of follow-up.
Complications
A total of fifty-two complications occurred in twenty-two patients. Only
five patients had no complications, sixteen patients had one complication, and
six patients had more than one complication. Of the major complications, the
most common, seen in seven patients, was a slow asymptomatic drift-through of
the superior rib cradle of the hybrid device through the rib of attachment,
with complete cutout, over a mean of three years (range, two to five years).
No skin erosion occurred superiorly because the lateral position of the
devices placed them under thick muscle. Three patients had two episodes of
migration. All devices were reseated, during the routine scheduled expansion
procedures, either to the initial rib of attachment (if it had reformed) or to
a more caudad or cephalad rib. In four patients, the spinal hook of the hybrid
device migrated posteroinferiorly, with complete disengagement at a mean of
two years (range, 1.5 to three years) but with minimal symptoms. These devices
were also reseated during scheduled expansion procedures.
Four patients had skin slough; three were treated with débridement
and primary closure, and one required temporary removal of the device and
later reinsertion. Four of five episodes occurred when the children became
more slender during a growth spurt. Two patients had one infection at the site
of the device, and one patient had two infections; all four infections were
associated with skin slough, and débridement and antibiotics were used
for treatment. The infection rate per surgical procedure was 1.9%. Six
patients had the development of a latex allergy. Three patients had mild
low-back pain, which resolved promptly after expansion of the hybrid device in
two of them. The third patient had symptomatic upper lumbar junctional
congenital kyphosis, which responded to a pedicle subtraction osteotomy. The
pediatric form of adult respiratory distress syndrome developed after the
surgery in two
patients14 and
required prolonged ventilator support before resolving without sequelae.
Upper-extremity brachioplexopathy developed in two patients after the surgery.
Both cases seemed to be related to malposition of the hybrid rib prosthesis,
and both resolved after repositioning. One nine-month-old boy sustained a
dural tear and spinal cord injury during resection of a thick mass of fused
ribs, and a Brown-Séquard syndrome developed. At the time of the last
follow-up, 2.8 years postoperatively, the patient had almost a full recovery.
Another patient, who had unilateral lung aplasia, survived a postoperative
episode of adult respiratory distress syndrome and did well clinically for
more than two years; however, respiratory failure eventually developed, and
the patient died following the development of pneumonia unrelated to surgery.
The remaining complications are summarized in the Appendix.
While it is well recognized that orthoses correct scoliosis indirectly
through pressure on the
ribs15, surgical
approaches based on this concept have been limited. In 1958,
Gruca16 described a
compression thoracoplasty in which indirect correction of idiopathic scoliosis
was obtained through longitudinal surgical compression of the ribs by
placement of steel springs on the convex side of the curve; however, problems
with dislodgment prevented the technique from being
accepted17. We
thought that, in young children, expansion thoracoplasty was more likely to
provide the necessary room for lung growth, so the opening wedge thoracostomy
was developed to directly address the constriction of the hemithorax resulting
from fused ribs and to indirectly address the congenital scoliosis without the
need for spine fusion.
The most favorable results with regard to the percentage of the predicted
normal vital capacity at the time of follow-up were seen in children who had
had the surgery at an age of less than two years, when lung growth by alveolar
cell multiplication is
greatest10,11.
Although it has not yet been proven that thoracic enlargement in humans can
improve lung growth, compensatory lung growth does occur after resection of
lung tissue in growing mammals, including humans, and this effect is greatest
in the very
young18,19.
Unilateral pneumonectomy in immature dogs provides a stimulus for the
remaining lung to double in size and in number of alveolar
cells20. At the
time of follow-up, patients who had been two years of age or older at the time
of the surgery had a lower percentage of predicted normal vital capacity than
did the younger patients and the three older patients with a history of spine
fusion had the lowest mean percentage of predicted normal vital capacity. This
is interesting in view of the fact that those three patients had better
correction of the scoliosis than did the patients who had been less than two
years of age and had more improvement in the lateral deviation of the spine as
indicated by the interpedicular line ratio. However, because there were only
three patients in the group, we cannot draw any firm conclusions. The space
available for the lung also increased in all groups, but this increase was not
significant in Group C, with the numbers available. The postoperative decrease
in the interpedicular line ratio was significant in all of the groups, whereas
the computed tomography indices were unchanged after treatment; thus, the
value of these measurements for the prediction of eventual vital capacity is
unclear. It should be added that the results of pulmonary function tests in
young children should be interpreted with
caution1.
Unrestrained growth of the thorax after correction of a three-dimensional
thoracic deformity early in life will probably best take advantage of the lung
growth curve10. The
percentage of the predicted normal vital capacity that is compatible with
long-term survival is controversial. Pehrsson et
al.21 suggested
that a vital capacity of <43% of the predicted normal value associated with
scoliosis is a risk factor for eventual respiratory failure. In our clinical
experience, we found that respiratory insufficiency may develop when the
predicted vital capacity is =25% of normal, but individuals with lower
values can do well
clinically22. A
normal individual loses a mean of 700 mL of vital capacity by the age of sixty
years23. This
natural loss of vital capacity with aging may explain why patients with
untreated severe infantile scoliosis, and probably with thoracic insufficiency
syndrome, who already have severe loss of vital capacity as a result of the
thoracic deformity begin to have an increased mortality rate (compared with
the normal rate) at the age of twenty years and have a rapid increase in this
rate after the age of forty
years21,24,25.
The loss of vital capacity with aging in patients with congenital scoliosis is
unknown.
General pulmonary health remained excellent for almost all of our patients,
with one having dramatic improvement but two others having deterioration with
the development of clinical respiratory insufficiency. Pulmonary infection may
also trigger respiratory insufficiency. One patient with unilateral lung
aplasia eventually had respiratory failure and died following development of
pneumonia unrelated to the surgery. We believe that all of the patients in
this series are at risk for the development of oxygen dependency, so although
those who remained on room air are considered to have had a successful
outcome, a much longer follow-up is needed to assess the possibility of future
oxygen dependency. To our knowledge, there is no series in the literature in
which patients with untreated severe congenital scoliosis were followed over
the long term to evaluate pulmonary function; thus, the natural history
remains undefined.
We were surprised that we were able to decrease the stiff congenital curves
in our patients by a mean of 25°. This is more correction than has been
seen following fusion, wedge resection, and convex epiphyseodesis
procedures26-29.
The advantage of our technique over those procedures is that the treated
portion of the curve, usually involving almost all of the thoracic spine, can
continue to grow with concurrent growth in the length of the thorax. The
scoliosis correction is equivalent to that described in reports on so-called
growing
rods3,30,
and most of the patients in those reports had less rigid, idiopathic or
neuromuscular, curves. In our patients, correction of the curve defined by the
Cobb angle and contained within the device span seems to be a combination of
correction of the central rigid segment of the curve with formation of small
curves in the opposite direction at each end of the rigid curve. The increase
in the length of the thoracic spine seen on radiographs was thought to be due
to postoperative growth, and this conclusion is supported by computed
tomographic analysis that also showed increases in the length of unilateral
unsegmented bars after expansion
thoracoplasty31.
Lack of growth of the thoracic spine may result in thoracic deformities like
Jarcho-Levin
syndrome22,32
with respiratory insufficiency due to extrinsic restrictive lung
disease1 from a
congenitally foreshortened thorax. Radiographic measurements showed that the
chests of our patients were growing, but those radiographic measurements
cannot be compared with the clinical chest dimensions of normal children
measured by Dimeglio and
Bonnel13 because
those authors performed external physical measurements with calipers. In
addition to growth and volume problems, the loss of thoracic symmetry due to
three-dimensional thoracic deformity is important because increasing asymmetry
of the thorax will likely adversely affect secondary breathing mechanisms and,
with a severe windswept deformity of the
chest1, actual
thoracic volume will be lost. Radiographs made in the coronal plane suggested
progression of the thoracic deformity before treatment and improvement
following treatment. In contrast, computed tomographic analysis in the
transverse plane suggested that the thoracic deformity was nonprogressive
before the surgery, but the mean preoperative survey time was only ten months,
so slow changes may have been missed. The thoracic deformity in the transverse
plane did not decrease significantly after treatment as hoped, but the
computed tomographic indices were stable during the treatment period. Whether
the deformity was inherently stable or the treatment played a role is
uncertain. While the fused hemithorax can be enlarged by our approach, it
cannot move with secondary breathing because of the absence of intercostal
muscles; thus, this procedure can address deficiencies in the thoracic
characteristic of stable
volume1 by
lengthening the hemithorax with osteotomy and distal hemidiaphragm transport,
but it cannot directly address deficiencies in the thoracic characteristic of
the ability to change
volume1.
The effect of spine surgery on long-term pulmonary function in children
with congenital scoliosis remains unclear. Does early surgical stabilization
of the spinal deformity result in a larger, more functional thorax at
maturity, or are the inhibition effects on the growth of the thoracic spine
conferred by surgery additive to those of the primary hemithoracic hypoplasia?
It is difficult to separate the effects of primary thoracic deformity from
those of early spinal surgery on lung development and function. In one study
of mostly older surgically treated patients with congenital
scoliosis33, those
with multiple thoracic anomalies had a mean vital capacity of 46% (range, 33%
to 59%) of the predicted normal value whereas patients with a similar degree
of scoliosis but only one or two spinal anomalies had a much better mean vital
capacity (mean, 74%; range, 59.4% to 88.6% of the predicted normal value).
Those with markedly compromised pulmonary function at the time of follow-up (a
vital capacity ranging from 27% to 46% of the predicted normal value) had four
to eleven anomalous vertebrae (mostly thoracic) and had undergone spinal
surgery at the mean age of 5.6 years (range, two to fourteen years). In
another report34,
patients with congenital scoliosis who underwent fusion at an age of 10.2 to
22.6 years (i.e., an age at which substantial thoracic growth had already
occurred) had a mean preoperative vital capacity of 71% of the predicted
normal value (range, 45% to 97% of the predicted normal value). The vital
capacity remained stable at the time of follow-up. O'Brien et
al.35 noted a low
vital capacity in very young children with severe spine and chest-wall
anomalies and a high mortality rate following surgical treatment. Although
infantile scoliosis has less of an adverse effect on vital capacity than does
congenital
scoliosis36,
Goldberg et al.37
recently reported that children with infantile scoliosis who had had spine
surgery before the age of ten years had a mean vital capacity of only 42% of
normal (range, 12% to 67% of normal) at maturity, whereas those who had had
surgery after the age of ten years had a mean vital capacity of 68% of normal
(range, 48% to 88% of normal). Winter and
Lonstein38 reported
on a patient with unilateral lung aplasia and congenital scoliosis who had
undergone posterior spine fusion at the age of three years. The patient had a
vital capacity of 70% of the predicted normal value at thirty-three years
postoperatively and had no clinical pulmonary problems at forty-one years
postoperatively. While the spinal deformity in these reported cases was well
characterized, the global thoracic deformity was not, so it is difficult to
compare our patients with those described in the literature. In summary, the
limited data available in the literature suggest that children with congenital
scoliosis are at risk for the development of restrictive lung disease,
especially if they have multiple levels of vertebral anomalies, but the effect
of spinal surgery in terms of altering that risk, either positively or
negatively, is not clear.
Our initial implant procedures are comparable with typical spine procedures
in magnitude, but the mean blood loss in our patients was less than that seen
with the subfascial rod
technique3. Our
patients had a mean of 1.8 procedures per year, but only the first surgery was
major. Although none of our patients would have been candidates for a
growing-rod procedure, comparisons between that technique and ours are useful
because of the similarities with regard to the approach and the need for
repetitive operations. The number of procedures per year in our patients is
comparable with that needed with the growing-rod
technique3. An
infection developed at the site of the device in 11% of our patients and
following 1.9% of the procedures, and it was always associated with a skin
slough. The rate of infection with the growing-rod technique has ranged from
7% (five of sixty-seven patients) to 18% (eight of forty-four patients), and
skin slough has been seen in 4% (three of sixty-seven patients) to 14% (six of
forty-four
patients)3,30.
The device migrated superiorly in 26% of our patients. The migration was not,
however, due to rib fracture; the rib cradle seems to migrate slowly upward
through the rib of attachment, often with the rib reforming caudad to it. The
device is then reattached, at the time of routine expansion, to the reformed
rib or to an adjacent one. In one series of sixty-seven patients treated with
the growing-rod
technique3, sixteen
(24%) had superior hook displacement and five (7%) had inferior hook
dislocation. Perhaps our dislodgment rate would have been lower if our
patients had worn a brace after the surgery, but we had concerns that bracing
would constrict growth of the chest wall.
Some patients treated early in the series had an injury of the brachial
plexus resulting from malposition of the device superiorly. The injury always
resolved after repositioning. Spinal cord injury occurred in one patient as a
result of an inadvertent violation of the spinal canal through an enlarged
neural foramen that had not been recognized on preoperative computed
tomography, but at 2.8 years postoperatively the monoplegia has almost
completely resolved. To our knowledge, no neurological complications have been
reported in association with the growing-rod
technique3,30,39,
but lower-extremity hyperesthesias have been seen with use of the so-called
Luque trolley40. In
a recent series of eleven patients treated with hemivertebrectomy, one had
transient lower-extremity weakness after the
surgery41. Spinal
cord monitoring and lower/upper extremity wake-up tests in the operating room
allow early recognition of these complications with the possibility of
immediate correction to minimize them. Patients with severe congenital
scoliosis often are frail and have comorbidities such as lung, heart, and
renal abnormalities. Although he was not included in this series, a
nine-month-old boy with congenital scoliosis and fused ribs and a history of
tracheoesophageal fistula died of bacterial pneumonia two weeks after the
surgery.
The weakness of our study is that the surgical procedure evolved over a
twelve-year period. Patients treated early in the series had less aggressive
correction, and five of them have had a second thoracostomy to gain more
correction. We think that the patients who were treated more recently had
better correction of the thoracic deformity, but long-term follow-up will be
needed to determine if the vital capacity trends remain favorable. We had no
controls, and the literature is limited with regard to studies of similar
patients and the reports that are available contain no analysis of the
thoracic deformity. Multicenter studies to address these issues are being
planned.
We agree with Dimeglio and
Bonnel13 that a
balanced thorax is just as important as a balanced spine since a balanced
thorax determines, to a great degree, long-term pulmonary function. Growth is
also important, and the thorax may be capable of more growth after correction
of the deformity. Opening wedge thoracostomy can correct most of the
three-dimensional thoracic deformity due to fused ribs and congenital
scoliosis, clearly decreasing the coronal and sagittal plane components of the
thoracic deformity and stabilizing the transverse plane deformity. The
scoliosis is stabilized without fusion, allowing spinal growth for increased
height of the thorax and additional gains in volume from growth of the rib
cage. At a minimum, use of this technique in small children with severe spine
and chest-wall deformities may delay the need for early definitive fusion
without compromising subsequent spinal procedures. Opening wedge thoracostomy
should not be used for a limited spine deformity, such as an isolated
hemivertebra, when the height of the thoracic spine is relatively normal for
the patient's age and the rib cage is normal in width and mobility, since
limited hemivertebrectomy or arthrodesis are more definitive procedures.
Opening wedge thoracostomy should be reserved for the most severe spine and
rib-cage deformities. We have found, however, that severe thoracic kyphosis
associated with the scoliosis is difficult to address with this approach.
Although the procedure currently requires frequent outpatient surgery for
expansion of the device, that will probably change in the future.
Self-expanding devices may be developed. Also, once the natural history and
outcomes of traditional treatment of children with severe thoracic
malformation are better understood, it might be possible to define a specific
age in childhood when an extensive thoracic spine fusion can be performed with
neutral or beneficial effects on long-term vital capacity, so that an end
point for treatment with opening wedge thoracostomy can be better defined.
In conclusion, opening wedge thoracostomy with use of a chest-wall
distractor directly treats segmental hemithoracic hypoplasia from fused ribs
associated with congenital scoliosis. The operation addresses thoracic
insufficiency syndrome by lengthening and expanding the constricted hemithorax
and allowing growth of the thoracic spine and the rib cage. The procedure
corrects most components of the thoracic deformity and indirectly corrects
congenital scoliosis in young children with an improvement in spinal alignment
and without the need for spine fusion. There probably is a benefit with regard
to the growth of the underlying lungs, especially in patients who are less
than two years of age at the time of surgery.
Note: The authors express their appreciation to Dr. G. Dean
MacEwen and Dr. John E. Hall for their encouragement in the early stages of
this study. Special thanks are also due to Mary Adams, Cindy Polasek, Bernice
Avilez, Lori Buegeler, RN, Randy Llamas, Jennifer Peunte, Mary Sueltenfuss,
RN, and Charmaine Grohman, RN. They appreciate the assistance of J. Walter
Simmons III, PhD, DO, and Earl Stanley, MD, in the validation trial for the
interpedicular line ratio measurements. They especially thank Michael Muhlert,
MD, for his assistance with the statistical analysis.