Scoliosis is a complex three-dimensional spinal deformity that most
often requires treatment to address curve progression during growth. Standard
treatment options for progressive scoliosis are essentially limited to bracing
or
surgery1-5.
While brace treatment is noninvasive and preserves growth, motion, and
function of the spine, it does not correct deformity and is only modestly
successful in preventing curve progression. Success of this form of treatment
may also be hampered by patient compliance issues and the negative
psychological impact of
bracing6-14.
In contrast, surgical treatment with an instrumented spinal arthrodesis
usually results in better deformity correction but is associated with
substantially greater risk. The risks of surgery are related to the
invasiveness of spinal arthrodesis, the instantaneous correction of spinal
deformity, and the profoundly altered biomechanics of the fused
spine15-22.
Fusionless scoliosis surgery may provide substantial advantages over both
bracing and
arthrodesis23,24.
The goal of this new technique is to harness the patient's inherent spinal
growth and redirect it to achieve correction, rather than progression, of the
deformity. Several terms for the treatment of scoliosis without arthrodesis
have evolved and imply different methods of fusionless scoliosis surgery.
These include endoscopic vertebral stapling, anterior spinal tethering, convex
scoliosis tethering, mechanical modulation of spinal growth, and internal
bracing of spinal
deformity23-25.
By applying implants directly to the spine, anterior fusionless techniques are
theoretically more advantageous mechanically than external bracing—as
bracing does not directly apply corrective forces to the spine but indirectly
transmits forces by means of the ribs, pelvis, and torso—and patient
compliance issues are eliminated. Furthermore, minimally invasive tethering of
the anterior thoracic spine by means of an endoscopic approach is also less
extensive than arthrodesis, with no requirement for discectomies, preparation
of the fusion bed, or harvest of bone graft.
All spinal implants that span a spinal motion segment are initially
subjected to a high stress that is usually intended to be
temporary26-30.
With the achievement of fusion, the high stresses are reduced or eliminated,
thus preventing instrumentation loosening or failure. Fusionless scoliosis
implants are not afforded the luxury of eventual fusion, and therefore issues
related to implant loosening or failure over time are important. There are few
data available on the longevity of fusionless scoliosis implants, and there
are no studies, as far as we know, that have compared different implant
strategies objectively with use of several indices of integrity in an in vivo
setting over time. To be effective over a prolonged period, these dynamically
loaded implants must maintain appropriate fixation to the host bone without
loosening or failure.
The purpose of this study was to compare the efficacy and integrity of more
rigid shape memory alloy staples and more flexible anterior thoracic tethers
for the treatment of progressive experimental scoliosis. Efficacy was defined
as the ability of the implant to control the progression of scoliosis. Plain
radiographs were used for in vivo assessment of efficacy. Integrity was
defined as the ability of the implant to maintain fixation in the host bone
during the treatment period. It was hypothesized that the use of a bone anchor
attached to a more flexible ligament loop tether would provide better control
of scoliosis progression than would the more rigid shape memory alloy staple.
It was also hypothesized that the use of a bone anchor attached to a flexible
ligament tether would demonstrate greater implant integrity.
Creation and Correction of Deformity
Under a study protocol approved by the Institutional Animal Care and
Use Committee, scoliosis was created in twenty-four Spanish Cross-X female
goats (six to eight weeks old and weighing 8 to 12 kg) with use of a flexible
left posterior asymmetric tether from the T5 to L1 laminae (3.5-mm
polyethylene core-polyester sleeve; Medtronic Sofamor Danek, Memphis,
Tennessee) as previously
described31. Convex
rib resection and concave rib-tethering from T8 to T13 were also performed to
create an experimental scoliosis (Fig.
1) similar to that described in our original
model32. After
eight weeks of posterior tethering, goats with progressive curves were
randomized into one of three treatment groups. Group I received no treatment
and served as scoliosis controls; Group II underwent anterior convex stapling
across the six levels of maximum curvature with rigid Nitinol shape memory
alloy staples (Medtronic Sofamor Danek); and Group III underwent anterior
tethering across the six levels of maximum curvature with flexible ligament
loops (3.5-mm polyethylene core-polyester sleeve; Medtronic Sofamor Danek)
attached to bone anchors (Medtronic Sofamor Danek). The goats were observed
for an additional twelve to sixteen weeks after treatment. The four-week range
in the observation period—necessitated by scheduling
constraints—was shared equally by all groups. An additional five goats
served as growth controls with no induced scoliosis. All animals were killed
at the end of the twenty to twenty-four-week total study period.
In Group II, shape memory alloy staples were placed anterolaterally along
the maximum curvature, with the staple base spanning the disc space vertically
and each tine anchored in adjacent vertebral bodies. Prior to implantation,
the staples were soaked in an ice bath causing them to become malleable. The
staple tines were then pulled to 90° angles for implantation. Once seated,
the staples were deployed by contact of the staple shoulder with an
electrocautery for approximately three seconds. This brief heating of the
staple shoulder initiated a bending of each staple tine to its former
"crimped" shape (a decrease in the staple tine-base angle from
90° at implantation to 70° to 80° after deployment).
In Group III, bone anchors were placed laterally along the maximum
curvature. Unlike placement of the shape memory alloy staples, anterolateral
placement of the bone anchors was not possible because of the length of the
anchor and the geometry of the goat vertebral body. Anterolateral placement of
the bone anchor would have resulted in violation of the spinal canal by the
tip of the implant. Additionally, whereas shape memory alloy staples were
placed without any preparation of the vertebral body, bone-anchor placement
required use of a trephine to core a path for the anchor. Bone from this core
was packed into the hollow chamber of the anchor prior to implantation. Once
in place, each adjacent pair of anchors was compressed and then an
appropriately sized ligament loop was positioned across the anchors to
maintain the tension in a corrected position. Mushroom-shaped caps were then
threaded into the anchors to prevent ligament dislodgement.
Radiographic Analyses
Serial plain radiographs in the posterior-anterior and lateral planes were
used to determine the magnitude of the deformity and the gross integrity of
the implants throughout the study. Progression of deformity was defined as an
increase in curve magnitude of 5° as measured with the Cobb
method33. All
staples and anchors that demonstrated evidence of loosening, including
radiolucency, drift, or back-out, were noted.
Histologic Analyses
After the goats were killed, the apical spinal segments (T9 to T11) were
harvested and placed in 70% ethanol to preserve the specimens before embedment
in polymethylmethacrylate with use of standard
techniques34,35.
Once the samples were completely embedded, a custom, water-cooled, high-speed,
cut-off saw36 was
used to section the samples with use of a diamond-edged blade (Rockazona,
Peoria, Arizona). Each section was then ground to a 2 to 3-mm thickness and
polished to an optical finish with a variable speed grinding wheel (Buehler,
Lake Bluff, Illinois) with use of standard
techniques37,38.
Sections were taken in the coronal plane through the vertebral bodies and
discs. Contact radiographs were made of each section and used for analysis.
Sections that provided the best representation of the center of the implant
and the vertebral body in the coronal plane were selected.
The sectioned vertebrae were initially qualitatively analyzed with use of
gross histologic techniques. Any observations regarding general implant
integrity were noted. Each 2-mm-thick section was then sputter-coated with a
conductive layer of gold for approximately two minutes (Hummer VI-A; Anatech,
Alexandria, Virginia) and was examined in a scanning electron microscope
(JSM-6100; JEOL, Peabody, Massachusetts) with use of the backscattered
electron detector (Tetra; Oxford Instruments, Concord, Massachusetts) to
provide a contrast between mineralized and nonmineralized tissues. Three
quantitative indices were obtained: an osseointegration index, a bone
proximity index, and the bone ingrowth.
The osseointegration index and bone proximity index were determined, with
use of an Image-Pro Plus software program (Media Cybernetics, Silver Spring,
Maryland), from the histological sections that best represented the center of
the implant. In Group II, two images allowed evaluation of the entire staple
while six images were required in Group III to capture the entire bone anchor.
The osseointegration index was calculated by measuring the length of the
implant surface osseointegrated with the surrounding bone and dividing this
value by the total length of the implant surface. The resultant percentage was
used to indicate the amount of bone osseointegrated with the implant. To
quantify the bone proximity index, trace lines were created over the implant
surface and the surrounding bone. The average distance between the two traces,
calculated by the Image-Pro software, represented the average distance between
the host bone and the implant surface. The bone ingrowth analysis was
accomplished in a manner similar to that used by Bloebaum et
al.39 for bone
ingrowth analysis within a porous coating. No bone ingrowth analysis was
possible for the staples, as these implants possess no internal chamber. For
each animal in Group III, three bone anchors (T9, T10, and T11) were analyzed.
Three images of the entire inner confines of the hollow chamber, as well as
two images of the host bone cephalad and caudad to the implant, were obtained
for each of the anchors. The host bone provided a normal percentage of bone
expected within a randomly sampled area of the vertebral body. Link ISIS
software (Oxford Instruments) was used to calculate the quantity of bone
within a specified area of each image. The percentage of bone within the
implant chamber was used to represent bone ingrowth and was compared for
reference to areas of surrounding bone.
Statistical analyses were performed on all of the histologic data with use
of independent t tests, with a level of significance defined as a p value
(alpha) of =0.05.
Implant Pullout Testing
After the goats were killed, the cadaveric specimens (less the apical
spinal segments T9 to T11, which had been used for histologic and
backscattered electron image analysis) were sealed in air-tight plastic bags
(S-1987; Uline, Waukegan, Illinois) and frozen to -20°C. Once thawed,
segments T7-T8 and T12-T13 were removed from the intact spinal columns and
were prepared for pullout testing. The strength of implant fixation was
analyzed at two time-points, the first representing the initial fixation
strength immediately after implantation (time zero) and the second
representing the final fixation strength at the end of the treatment period
(twelve to sixteen weeks after implantation). Vertebral bodies from the
untreated animals (Group I and growth controls) were used to generate
time-zero data, whereas the treated animals (Groups II and III) were used to
generate twelve to sixteen-week data. Specimens used for time-zero pullout
testing were cleaned of all soft tissue to expose the bone of the vertebral
bodies. Specimens used for the twelve to sixteen-week pullout testing
underwent a minimal amount of soft-tissue dissection to allow for application
of the pullout fixtures to the implants. For the staples, a sewing needle was
used to guide a spider wire through the fibrous tissue and under each shoulder
of the staple (Fig. 2). For the
bone anchors, fibrous tissue surrounding the exposed portion of the anchor was
removed (Fig. 3). Following
dissection, specimens were coated with a thin layer of a water-based lubricant
to minimize dehydration and were kept moist with use of periodic sprays of
0.9% saline solution. Though staples likely resumed a malleable crystal
structure with freezing, no manipulation of the staple occurred, and therefore
the overall shape of the staple in the bone was unchanged. Heating of all
staples with use of an electrocautery prior to pullout testing merely ensured
that the shape memory alloy was in its deployed crystal structure state at the
time of pullout testing.
Pullout testing to generate time-zero data was performed on the six
untreated scoliotic spines as well as the five growth control goats. Staples
were implanted across the T7-T8 and T12-T13 segments in these eleven goats
post mortem. After staple testing, bone-anchor pullout testing was
accomplished in the same four vertebrae. Because of the small diameter of the
tines, staple implantation and pullout testing resulted in a 3-mm cylindrical
hole in the lateral vertebral body with minimal change in the surrounding
cancellous bone. Subsequent preparation for implantation of the 8-mm-diameter
bone anchor involved the use of a 6-mm trephine to core a path for the bone
anchor around the previous staple path. This method of bone-anchor
implantation essentially mimicked that used during the in vivo portion of the
study. This procedure was used to conserve the limited supply of specimens and
to increase the power of the statistical analyses. The T7-T8 and T12-T13
vertebrae from the fourteen treated goats were used to determine the pullout
strengths at twelve to sixteen weeks after implantation.
Implant pullout testing was performed with use of a servohydraulic
materials testing machine (model 8500; Instron, Canton, Massachusetts).
Testing of the implanted devices was performed by securely bracing the
implanted specimen in a specially designed jig such that the implanted device
was aligned with the Instron actuator, allowing for a tensile load to extract
the implant. A 5-N preload was placed on each specimen followed by a constant
displacement rate of 1 mm/sec until failure occurred.
Implant pullout data were analyzed with use of a one-way analysis of
variance followed by a Tukey-Kramer HSD (honestly significant difference)
post-hoc test. Statistical significance was defined as a p value (alpha) of
=0.05.
Radiographic Analyses
Of the twenty-four goats that underwent posterior asymmetric
tethering with convex rib resection and concave rib-tethering, two (8%) died
in the postoperative period because of pulmonary complications. Of the
twenty-two goats available for analysis at the end of the eight-week tethering
period, twenty (91% of the surviving animals; 83% of all study animals) had
progressive, structural, idiopathic-type, lordoscoliotic curves develop convex
to the right in the thoracic spine. All twenty goats with progressive
scoliotic curves demonstrated radiographic and clinical features
characteristic of idiopathic scoliosis. Radiographically, these features
included substantial displacement of the apical vertebrae from the midline;
wedging of the apical vertebral bodies and discs; and rotation of the apical
vertebra, with a grade of 2 or 3 according to the criteria of Nash and
Moe40. Clinically,
these features included decreased flexibility of the spine (determined with
use of a push prone maneuver); and, after rib regeneration, a typical,
idiopathic-type, posterior thoracic deformity involving a right rib prominence
and flattened left thoracic cage.
Over the eight-week tethering period, curves progressed from an average
(and standard deviation) of 57.2° ± 8.3° to 76.5° ±
9.3° in the coronal plane (p < 0.001) and from -18.9° ±
3.8° to -40.9° ± 7.6° in the sagittal plane (p < 0.001).
The goats were then randomized into the three treatment groups, at which point
no significant differences in curvature were demonstrated in the coronal or
sagittal planes between any of the three groups. During the treatment period,
the scoliosis progressed in the seven goats treated with staples (Group II)
from an average of 77.3° ± 11.5° to 94.3° ±
12.2° (p < 0.05) (Figs. 4-A, 4-B,
and 4-C), demonstrating no significant difference in progression
of scoliosis (p = 0.90) compared with the six untreated goats (Group I), in
which the scoliosis progressed from an average of 79.5° ± 7.6°
to 96.8° ± 6.7° (p < 0.05)
(Table I). In contrast, the
scoliosis in the seven goats with ligament tethers attached to bone anchors
(Group III) corrected from an average of 73.4° ± 8.4° to
69.9° ± 9.7° (p = 0.34)
(Figs. 5-A, 5-B, and 5-C).
Qualitatively, serial radiographs demonstrated progressive loosening of
eighteen (43%) of forty-two staples, with two staples becoming completely
dislodged. Only two (4%) of forty-nine anchors demonstrated a slight drift
without radiolucency.
At the beginning of the treatment period, the sagittal plane deformity in
the untreated group measured an average of -40.3° ± 6.3° and
progressed to -61.0° ± 18.8° of lordosis (p < 0.05) over
twelve to sixteen weeks. The sagittal plane deformity progressed from an
average of -37.3° ± 6.6° to -49.0° ± 14.7° of
lordosis (p = 0.03) in the group treated with staples and from -44.4° to
-58.9° of lordosis (p < 0.003) in the group treated with bone anchors
and ligament tethers over the twelve to sixteen-week treatment period.
Histologic Analyses
Gross inspection of the histologic sections revealed a consistent halo of
fibrous tissue around the staple tines but intimate contact of the surrounding
host bone with the bone anchors at all sites, including the two anchors that
demonstrated slight drift (Fig.
6). This finding was consistent with the measurements collected
from the backscattered electron images
(Fig. 7). The average
osseointegration index value (and standard error) for the seven goats treated
with staples (40.1% ± 17.5%) was significantly less than that
calculated for the seven goats treated with bone anchors (76.4% ± 6.8%)
(p < 0.001). The average bone proximity index was 702.0 ± 138.0
µm for the staples and 215.0 ± 69.0 µm for the bone anchors; the
difference was significant (p < 0.001). In addition, there was an average
of 11.0% ± 3.8% bone ingrowth within the hollow chamber of the bone
anchor compared with 21.7% ± 4.9% bone in the host region (p =
0.004).
Implant Pullout Testing
The average staple pullout strength was 101.4 ± 23.0 N for
twenty-two staples at time zero and 86.0 ± 48.6 N for thirteen staples
at twelve to sixteen weeks after implantation (p = 0.3). (One staple was
dislodged and therefore in a suboptimal position for implant pullout testing.)
The average bone-anchor pullout strength was 495.4 ± 171.3 N for
forty-four anchors at time zero and 639.8 ± 213.4 N for twenty-eight
anchors at twelve to sixteen weeks after implantation; the increase was
significant (p = 0.004). At both time-points, the bone-anchor pullout strength
was significantly greater than that of the staples (p < 0.001)
(Fig. 8). The mode of failure
differed between staples and anchors. At twelve to sixteen weeks after
implantation of the devices, pullout testing resulted in failure of the
vertebral body bone for twenty-one (75%) of the twenty-eight anchors, whereas
all staples pulled out cleanly with no host-bone failure.
Although the term fusionless scoliosis surgery is currently used to
describe definitive anterior spinal procedures that control the progression of
scoliosis during growth, other fusionless treatments have been in existence
for years. These more established fusionless procedures most often use
posterior implants to control the progression of spinal deformity in younger
children41,42.
However, fusionless scoliosis surgery with use of anterior implants provides
theoretical advantages over posterior procedures. Subcutaneous or submuscular
rod techniques and the vertical expandable prosthetic titanium
rib41 are not only
potentially more invasive than fusionless scoliosis surgery but may be
associated with an increased rate of complications. Additionally, the
procedures are temporizing and require multiple surgeries throughout growth
with the ultimate goal of spinal fusion. Fusionless scoliosis surgery avoids
multiple procedures, as well as the requirement for an eventual arthrodesis,
by offering a single intervention that may provide a more permanent solution
to the spinal
deformity23,43.
Furthermore, substantial correction of a spinal deformity in the absence of a
rigid fusion mass spanning several vertebral motion segments may prevent some
of the long-term problems related to spinal arthrodesis with instrumentation.
These include altered stress on adjacent unfused segments and spinal imbalance
issues16-22,44.
The data from this study demonstrate greater efficacy and integrity of a
bone anchor attached to a more flexible ligament loop tether compared with a
more rigid shape memory alloy staple in the fusionless treatment of a
progressive experimental scoliosis. The greater efficacy of the bone anchor
attached to a ligament tether in controlling scoliosis progression was
demonstrated in vivo over the course of the twelve to sixteen-week treatment
period, and implant pullout testing demonstrated superior fixation of the bone
anchor both at the time of initial implantation (time zero) and at the end of
the treatment period (twelve to sixteen weeks after implantation). In contrast
to the more rigid staple base, the ligament loop used with the bone anchor
provided a more flexible tether spanning the disc space that was likely
associated with decreased forces during spinal motion. This potentially
protected the bone anchor from loosening over the course of the study.
The pullout testing was important not only in highlighting differences
between the two implants at given time-points but also in demonstrating
changes in integrity within an implant over the course of the study. Whereas
the staple demonstrated no significant change in pullout strength between the
two time-points, the bone anchor showed a significant increase in pullout
strength. It is speculated that this difference was related to two factors:
(1) the rigidity of the portion of the implant spanning the motion segment,
and (2) the quality of the fixation to bone. The staple, though made of shape
memory alloy, has a relatively rigid base spanning the disc space compared
with the ligament loop-bone anchor construct. For a given displacement across
the disc space, created by the motion of the spine, greater forces were likely
generated at the junction of the implant and host bone in the more rigid
staple. These higher forces perhaps contributed to the increased loosening and
the trend toward decreased fixation strength of the staple over the course of
the study. Additionally, the smooth tine is suboptimal for fixation to bone
and relies primarily on the mechanical "crimping" effect of the
deployed shape memory alloy staple.
While different methods of implant pullout testing are possible, axial
pullout was chosen to allow a simple comparison of a limited number of
specimens. Although the axial pullout testing performed at time zero in this
study did not approximate the mode of failure for these implants in vivo, it
did demonstrate the substantial difference in the initial strength of osseous
fixation between the implants. The pullout testing at twelve to sixteen weeks
may have provided a more appropriate estimation of the in vivo strength of the
implant, as these implants were subjected to physiologic spinal loads in live
goats over an extended period of time.
It is possible that the greater efficacy of the bone anchor attached to a
flexible ligament loop tether compared with a rigid shape memory alloy staple
in controlling scoliosis progression was due in part to the superior integrity
of this implant over the course of this study. However, the initial scoliosis
correction of 11.4° achieved by this device compared with the initial
correction of 1.2° achieved with the staple likely altered the
biomechanical environment in favor of the bone anchor-ligament loop
construct.
Despite the superior performance of the bone anchor attached to a ligament
loop tether in this experimental scoliosis animal model, some caution is
appropriate in making comparisons with the treatment of human idiopathic
scoliosis. Our model, although it approximates idiopathic scoliosis, does not
mimic this condition. Indeed, without a clearly defined
etiology45, it is
impossible to reproduce a true idiopathic scoliosis. However, our previous
work32 has
suggested similarities to idiopathic scoliosis clinically, radiographically,
histologically, and biochemically. The scoliotic deformities created in this
study were of an extreme magnitude and demonstrated a malignancy of
progression that is not commonly seen in idiopathic scoliosis. Yet, the
mechanical factors related to progression of scoliosis with growth, according
to the Hueter-Volkmann
principle46-50,
are well simulated in this model. The experimental scoliosis created in this
study represents a challenging scenario for the testing of fusionless
implants.
Other limitations of this experimental scoliosis model are apparent when
contrasting the biomechanics and anatomy of animals and
humans32. The
postural differences between a quadruped goat and a bipedal human likely
create forces on the spine that are not directly comparable. The anatomy of
the goat thorax is also more pyramidally shaped and stiffer than the cubical
human thorax. However, there are many similarities between the goat thorax and
the human thoracic spine, making the goat thorax a reasonable enough
approximation of a juvenile human spine that it can provide useful information
for the study of progressive scoliosis and its treatment.
Previous attempts to correct scoliosis with anterior fusionless techniques
have been
disappointing51,52.
Nachlas and
Borden52 were
initially optimistic about their ability to create and correct lumbar
scoliosis in a canine model using a (rather weak) staple spanning several
vertebral segments. The enthusiasm for this new treatment waned after the
application of their staple in three children with progressive scoliosis met
with poor results. Other investigators in the
past51 have also
been dissatisfied with convex stapling as a means of controlling progressive
scoliosis.
More recent investigations of convex vertebral body stapling, both in
animal models and in juvenile and adolescent scoliosis, however, have offered
promising early results with use of improved implants and
techniques23,25,43.
The use of a shape memory alloy staple tailored to the size of the vertebral
body, the application of several staples per level, the instrumentation of all
levels of curvature, and the employment of minimally invasive endoscopic
approaches all offer substantial improvements over previous fusionless
techniques. Patient selection may also play a role in the current success of
these fusionless treatments, with perhaps the ideal candidates for this
intervention possessing smaller and more flexible single thoracic curves. Yet,
with the early clinical success of these stapling procedures, no basic-science
data are available to assist in the evaluation of these implants and their
effect on the surrounding tissues.
The model used in this study provides a unique environment for the
evaluation of novel fusionless techniques with use of objective radiographic,
histological, and biomechanical analyses to compare various strategies.
Improvements in implant design in this experimental model, with a specific
focus on optimizing the fixation to bone and maximizing the tethering effect,
may lead to greater control of idiopathic scoliosis in children. ?
Note: The authors would like to thank Michelle Swenson for her
technical assistance with the collection of the data.