There were approximately seven million fractures in the United States in
19981. Two-thirds of
these fractures were closed, and the majority of them were treated with closed
reduction. About 8.6% of all fractures occurring annually are
tibial1. The average
healing time of tibial fractures treated with an intramedullary nail is
approximately seventeen to twenty-two
weeks2. Long
rehabilitation times with associated loss of work hours are often required
before the patient can return to full function. The prevalence of
complications associated with tibial fractures has ranged from 5% to
20%1. These
complications include compartment syndrome, delayed union, nonunion, and
malunion3-5.
These findings indicate that considerable clinical benefit would be derived
from accelerating tibial fracture healing as well as fracture-healing in
general and by avoiding the secondary procedures required to treat delayed
unions and nonunions.
A number of osteogenic factors have been associated with accelerated
bone-healing in animal models and human clinical
trials6,7.
Recombinant human bone morphogenetic protein-2 (rhBMP-2) and 7 (OP-1) are
differentiation factors with bone-inducing
properties8,9.
Unlike other factors, these osteogenic BMPs are capable of initiating the
entire cascade of de novo bone
formation10-12.
Although rhBMP-2 and 7 delivered in solvent buffer alone have been shown to
accelerate bone-healing in a number of animal models, the use of other
delivery vehicles can enhance their therapeutic
effect12-16.
A major role of delivery vehicles, especially in large animal fracture models,
is to maintain the concentration of osteogenic factors at the fracture site
for a sufficient period of time to induce bone-forming cells to migrate to the
area of injury, proliferate, and differentiate. Delivery vehicles for
substances that promote fracture-healing must be porous and rapidly
biodegradable to allow cellular and vascular invasion while minimizing the
effects of residual carrier on the biomechanical properties of the repair.
Ideal carriers should also be biocompatible in order to limit interference
with bone induction caused by an excessive inflammatory response. In addition,
delivery vehicles for closed fracture repair should be injectable through a 16
or 18-gauge needle to minimize discomfort to patients during percutaneous
administration.
Although many of the delivery vehicles for administration of osteogenic
factors require open surgical
implantation16,17,
several carriers, including calcium phosphate cements, have been investigated
for their suitability for percutaneous
delivery15-19.
Calcium phosphate cements that set under endothermic conditions and form
poorly crystalline hydroxyapatite are ideally suited for percutaneous delivery
of osteogenic
factors20,21.
The endothermic setting reaction eliminates thermal damage to proteins.
Calcium phosphate cements also have a high affinity for binding proteins such
as osteogenic agents. Poorly crystalline hydroxyapatites are resorbed quickly
by osteoclasts and giant cells, minimizing interference with fracture-healing
while they release the osteogenic
agents20-23.
Since the mineral phase of bone is composed primarily of hydroxyapatite,
osteoblasts can deposit bone directly onto these osteoconductive carrier
materials. In addition, these carriers are radiopaque and can easily be
identified during injection by fluoroscopy.
The purpose of this study was to evaluate the capacity of a single
percutaneous injection of rhBMP-2 combined with a-BSM to accelerate
healing of a canine mid-diaphyseal tibial osteotomy site.
Experimental Design
Bilateral mid-diaphyseal tibial osteotomy was carried out in sixteen adult
female purpose-bred dogs (mean weight, 22.9 kg; range, 18.5 to 27.5 kg), and
the sites were stabilized with external fixators. Four hours after the
surgery, one limb of each dog was randomly treated either with a percutaneous
injection of 1.5 mL of rhBMP-2/a-BSM paste containing 1.05 mg of rhBMP-2
or with a percutaneous injection of an equal volume (1.5 mL) of a-BSM
alone. The type of treatment was determined with a randomized block design.
There were eight limbs in each group, and the contralateral limb was not
treated with any substance after the osteotomy, to serve as a control. Eight
of the contralateral limbs had a sham insertion of a needle and manipulation
without injection of material. Results were evaluated with serial in vivo
radiography and force-plate analysis. In addition, evaluations of mechanical
properties and histologic measurements were carried out on excised tibiae
obtained eight weeks after treatment. The university animal care and use
committee approved all procedures.
rhBMP-2/a-BSM and a-BSM Formulations
rhBMP-2 was reconstituted with the appropriate volume of pH 4.5 buffer (5.0
mM L-glutamic acid, 5.0 mM NaCl, 2.5% glycine, 0.5% sucrose, and 0.01%
polysorbate 80). The rhBMP-2/a-BSM paste was formulated by mixing 2.5 mL
of a 0.9-mg/mL rhBMP-2 solution injected with 2.5 g of a-BSM powder (1.0
liquid-to-powder ratio) in the supplied silicone mixing bulb for 1.5 minutes.
The a-BSM paste was formulated by adding 2.5 mL of buffer to 2.5 g of
a-BSM powder. The a-BSM powder was made of a blend of two calcium
phosphate precursor powders, amorphous calcium phosphate and dicalcium
phosphate dihydrate. This combination yielded 3.25 mL of 0.70-mg/mL
rhBMP-2/a-BSM paste. The 0.70-mg/mL rhBMP-2/a-BSM concentration
was used on the basis of a study of a rabbit osteotomy model treated with the
same rhBMP-2/carrier
combination24 and a
study of a canine allograft model in which rhBMP-2 was delivered in an
absorbable collagen
sponge25. The
appropriate volumes of each paste were then transferred to 3-mL syringes.
Injections were performed through an 18-gauge spinal needle within five to ten
minutes following preparation. The force required for injecting the paste with
this syringe-and-needle combination was approximately 20 N. The paste
formulations were injectable for at least thirty minutes. The time for
complete hardening at body temperature was less than one hour. Following
injection into the osteotomy hematoma four hours after the surgery, both
pastes formed cement granules within and around the osteotomy site.
Surgical Procedures
All dogs were premedicated with butorphanol and acepromazine
intramuscularly, anesthetized with an intravenous injection of thiopental, and
intubated, and the anesthesia was maintained with halothane in 100% oxygen. A
lumbosacral epidural injection of bupivacaine and morphine was performed prior
to surgical clipping and aseptic preparation to supplement the anesthesia and
to provide immediate postoperative analgesia. In addition, two 100-µg/hr
fentanyl patches were applied to the lateral aspect of the thorax. Bilateral,
craniolaterally placed unilateral external fixators (Orthofix, Richardson,
Texas) were
applied26. A 5-cm
skin incision that extended to the periosteum was then made on the
craniomedial aspect of the leg. The periosteum was bluntly elevated
circumferentially, and a 1-mm gap was created bilaterally by performing a
mid-diaphyseal tibial osteotomy with an oscillating saw and copious
saline-solution lavage. The side bar of the fixator was then removed, and the
fibula was fractured at the level of the tibial osteotomy. A 1-cm section of
the fibula was removed both proximal and distal to the fracture with rongeurs.
The side bar was then replaced, and cortical alignment and the width of the
osteotomy gap were confirmed with a 1-mm spacer. The fixators were secured,
the incisions were closed in three layers, and the dog was allowed to recover
from the anesthesia.
Four hours after creation of the osteotomies, the dog was reanesthetized
and one limb was randomly treated with a fluoroscopically guided percutaneous
injection of rhBMP-2/a-BSM or an equal volume of a-BSM alone. The
18-gauge, 3.8-cm needle was inserted through the craniolateral muscle mass
into the osteotomy site with fluoroscopic guidance. The needle was directed
across the osteotomy site, and orthogonal imaging was used to verify the
location of the needle next to the cortex opposite the needle entry site.
Approximately 0.5 mL of the appropriate treatment paste was injected at the
periosteal-soft tissue interface at the surface of the far cortex. The needle
was then withdrawn to the center of the medullary space, where an additional
0.5-mL volume was injected. The needle was withdrawn again, to the near
cortex, where the final 0.5-mL volume of paste was injected at the
periosteal-soft tissue interface. The 1.5-mL volume of material had been
found, in pilot trials, to adequately cover the 1.0-mm osteotomy site. The
osteotomy site of the contralateral limb was randomly chosen to be left
untreated or to be subjected to a sham needle insertion and manipulation. All
dogs received 500 mg of oral tetracycline daily for continuous new-bone
labeling until the animal was killed, eight weeks after the surgery.
Radiography
Anteroposterior and mediolateral radiographs were made before surgery,
immediately after stabilization of the osteotomy site, immediately after
injection of the rhBMP-2/a-BSM or a-BSM, and at four and eight
weeks following surgery. Union of the osteotomy site was graded on radiographs
by three independent observers using a scale of 1 to 4, with 1 indicating
poor, 2 indicating fair, 3 indicating good, and 4 indicating excellent
(Table
I)25.
The initial distribution of the a-BSM carrier and the subsequent area
(in square millimeters) of the periosteal callus were measured at each cortex
(medial, lateral, anterior, and posterior), over a 15-mm length proximal and
distal to the original osteotomy site, with use of a digitizer and specialized
software (NIH Image; National Institutes of Health, Bethesda,
Maryland)26. The
total area was calculated as the sum of the areas of the callus over the four
cortices. Radiographs at four and eight weeks after the surgery were compared
with immediate postoperative, postinjection radiographs, and the percentage of
residual carrier relative to the amount of carrier present immediately after
the injection was scored subjectively by three independent observers on a
scale of 0 to 4 (Table I), with
0 indicating no residual material remaining and 4 indicating >75% residual
material remaining. Residual a-BSM could be distinguished from newly
formed bone by differences in density and the lack of a trabecular
pattern.
Force-Plate Measurements
Weight-bearing with the external fixators in place was measured dynamically
while the dogs trotted at a standard velocity (2.5 m/sec) over a force
platform (ORS6-6-1000; AMTI, Watertown, Massachusetts) before the surgery and
at four and eight weeks after the surgery. Time-integrated vertical force
(N-sec/kg), which represents the total amount of vertical force for each
hindlimb during one complete stride, was normalized to body weight (kg). The
mean normalized time-integrated vertical force for each hindlimb was
determined from three or more valid foot strikes on the platform with use of a
commercial software program (Acquire, version 6.21W; Sharon Software, DeWitt,
Michigan)27. Foot
strikes were considered to be valid when the dog's velocity did not change by
more than 10% leading up to the force-plate measurement.
Mechanical Testing
Mechanical testing was performed within two hours after the harvest of the
specimen. Prior to testing, the external fixators were removed and the tibiae
were transected between the most proximal and distal pairs of pins, resulting
in a specimen of 12 cm in length. Two 3-mm-diameter pins were placed
perpendicular to each other at both the proximal and the distal extent of the
excised specimens. The ends of the specimen were then potted in
polyester/styrene with an alignment jig for testing. The span length of the
exposed part of the specimen was 7 cm. Mechanical testing was performed on a
modified servohydraulic materials testing system (model 858; MTS Systems, Eden
Prairie,
Minnesota)28.
Continuous recording of the load versus deformation was made during the
testing.
The specimens were first tested nondestructively in compression,
anteroposterior and mediolateral bending, and torsion with use of flex control
grips28,29.
Five separate load control cycles at 0.2 Hz were conducted for each
nondestructive test. Axial compression was cycled from 0 to 50 N in
compression. Anteroposterior and mediolateral bending were cycled with a
sinusoidal ±2.0-N-m moment. Torsion was cycled at 1.5°/sec to a
maximum of 5° of displacement with use of ±2.5-N-m torque. The
mounting grips allowed the MTS load frame to produce pure axial load or torque
about the long axis of the bone. In bending tests, torque cells connected
electric motors to the bone ends. The motors created a constant bending moment
over the length of the tibia with axial load and torsion maintained at zero
under load control. The reproducibility of the nondestructive testing data was
confirmed by repeatability within the five cycles tested.
Following nondestructive testing, the tibiae were tested to failure in
torsion in external rotation at 1.5°/sec to a maximum of 45° or until
failure. Load and deformation data were recorded continuously at 10 Hz with an
A/D board and stored on a personal computer in a data file. Stiffness was
calculated as the slope of the initial linear portion of each curve. Ultimate
torque, angular deformation at failure, and total energy to failure were also
determined from the data for each specimen. The location of the failure was
recorded as being through the original osteotomy site or outside the osteotomy
site and through normal diaphyseal
bone30.
Histologic Analysis
The mechanical testing apparatus returned the specimens to the original
starting point and allowed alignment of the failed specimens for histologic
processing. Immediately after the completion of the mechanical testing, the
specimens were fixed in 70% ethanol for histologic processing. The bones were
processed for calcified histologic
analysis31. The
tissue was dehydrated and was embedded in methylmethacrylate. Three sections
from each block were cut frontally to a width of 200-µm with use of a
diamond-wafering blade (Isomet 2000 Precision Saw; Buehler, Lake Bluff,
Illinois). Each section was subsequently ground to 80 to 100 µm with use of
a speed-lapping machine (ML-521D; Maruto Instrument, Tokyo, Japan).
Fine-detail contact microradiography was performed (18 kVp, two minutes) with
a Faxitron apparatus (model 43855A; Hewlett-Packard, McMinnville, Oregon) on
one section from each limb. These sections of bone were stained with a
modified Goldner trichrome method for calcified bone. The slides were
evaluated qualitatively for callus composition and maturity. Bridging of the
osteotomy site and cortical continuity were recorded for each tibia. The total
callus area and the area of fibrous tissue, cartilage, bone, and residual
carrier were measured by tracing the corresponding regions within the callus
with use of NIH image
software26.
Histologic slides were evaluated at 20× and 40× magnification for
the percentage of residual a-BSM relative to the total callus area.
Residual a-BSM could be easily distinguished from normal bone by its
granular appearance, increased density, and lack of cells.
Data Analysis
The mean and standard error of the mean were determined for each
measurement. Statistical analysis was performed with the general linear models
technique for analysis of variance (SAS Release 7.01; SAS Institute, Cary,
North Carolina). Paired nonparametric comparisons of the radiographic union
scores (rhBMP-2-treated group compared with their contralateral controls and
a-BSM-treated group compared with their contralateral controls) and of
a-BSM-retention grades (rhBMP-2-treated group compared with
a-BSM-treated group) were carried out with use of the Wilcoxon
signed-rank test. Unpaired nonparametric comparisons (rhBMP-2-treated group
compared with a-BSM-treated group and the control limbs of the
rhBMP-2-treated dogs compared with the control limbs of the
a-BSM-treated dogs) and comparisons of radiographic union at four and
eight weeks were made with use of the Mann-Whitney U test. Interobserver
precision errors were determined for the analyses of the radiographic union
scores and the a-BSM-retention grades. Paired parametric comparisons of
the mechanical testing data (rhBMP-2-treated group compared with their
contralateral controls and a-BSM-treated group compared with their
contralateral controls) were performed with use of the Student paired t test.
Unpaired parametric comparisons (rhBMP-2-treated group compared with
a-BSM-treated group and the control limbs of the rhBMP-2-treated dogs
compared with the control limbs of the a-BSM-treated dogs) were made
with use of unpaired t tests. Time series data (callus area and findings of
force-plate analysis) were evaluated with repeated-measures analysis of
variance. When significance was detected, differences were separated with use
of the Duncan multiple-range test. The locations of the failure during
mechanical testing were also compared with use of the Fisher exact test. All
of the comparisons were two-tailed, and significance was always set at p <
0.05. When comparisons were not significant, the difference between
populations necessary to detect a significant difference with a power of 0.8
and a= 0.05 was calculated.
No significant differences with regard to any of the variables were
observed between the contralateral, control limbs left untreated after the
osteotomy and the control limbs that received a sham needle insertion and
manipulation. Therefore, all of the subsequent analyses involved comparison of
the rhBMP-2/a-BSM group and all eight of its contralateral controls or
the a-BSM group and all eight of its contralateral controls.
The percutaneous injection of rhBMP-2/a-BSM was easily accomplished
four hours after the surgery. There was no extravasation of the paste through
the surgical wound and no apparent ectopic mineralization of soft tissues
along the needle tract or in the tibial compartments adjacent to the healing
osteotomy callus. With the exception of mild postoperative swelling, all of
the dogs recovered uneventfully from the surgery and were walking within five
days postoperatively. At four weeks after the surgery, six of the eight dogs
in the rhBMP-2 group and five of the eight in the a-BSM group were not
visibly lame. The mild lameness of the remaining two dogs in the rhBMP-2 group
was associated with the untreated, control limb. Two of the dogs in the
a-BSM group were mildly lame because of the control limb at four weeks,
and the remaining dog had mild lameness associated with the
a-BSM-treated limb. None of the dogs in the rhBMP-2 group were lame at
eight weeks, whereas one dog in the a-BSM group was mildly lame at eight
weeks because of the control limb.
Distribution of a-BSM, Radiographic Healing, and Callus
Area
Radiographic evaluation confirmed the localization of the radiodense
a-BSM carrier around the osteotomy site in both the
rhBMP-2/a-BSM-treated and the a-BSM-treated limbs after injection
(Figs. 1, A, and
2, A). Radiographs of
the rhBMP-2-treated limbs made at four weeks demonstrated the dispersion of
particles of the injected material (Fig. 3,
B). In addition, there was considerable new bone
formation in and around the residual radiodense a-BSM carrier at four
weeks. At eight weeks after the surgery, the original osteotomy site was less
visible and there was remodeling of the bridging callus
(Fig. 1, C). Initial
callus formation was apparent in the contralateral, untreated control limbs of
the rhBMP-2-treated dogs at four weeks
(Fig. 1, E). At eight
weeks, the site of the control osteotomy was still visible and the callus had
not fully bridged the site on all four cortices
(Fig. 1, F).
Radiographs of the a-BSM-treated limbs made at four weeks demonstrated
much less particulation and dispersion of the a-BSM as well as less bone
formation than did the radiographs of the rhBMP-2-treated limbs made at the
same time point (Figs. 3, B, and
3, D). At eight weeks, the osteotomy site was still
visible and there was little evidence of bridging callus
(Fig. 2, C). Residual
radiodense material presumed to be a-BSM was also still present. The
radiographic appearance of the untreated, contralateral control limbs of the
a-BSM-treated dogs (Fig. 2, D,
E, and F) was similar to that of the control limbs
of the rhBMP-2-treated dogs.
The callus area in the rhBMP-2-treated limbs was significantly greater than
that in the a-BSM-treated limbs at four and eight weeks after the
surgery (Table II). The callus
area in the rhBMP-2 group at eight weeks was decreased compared with the area
at four weeks (p < 0.05). The callus area in the a-BSM-treated limbs
was greater than that in the contralateral, untreated control limbs at four
weeks (p < 0.05) but not at eight weeks (power = 0.8 to detect a difference
of 28%). The callus area of the control limbs of the rhBMP-2-treated dogs and
the control limbs of the a-BSM-treated dogs increased from four to eight
weeks (p < 0.05). There was no difference in the callus area between the
two control groups at either time point (power = 0.8 to detect a difference of
31% at four weeks and power = 0.8 to detect a difference of 36% at eight
weeks). Thus, there was a significant group effect (p = 0.01), time effect (p
= 0.0001), and group × time interaction (p = 0.0001) for callus
area.
Radiographic Analysis (Scores for Union of Osteotomy Site and
Residual a-BSM)
The radiographic union scores were significantly greater for the
rhBMP-2-treated limbs than they were for the a-BSM-treated limbs and for
both contralateral, untreated control groups at four weeks and eight weeks
(Table III). All union scores
increased significantly from four to eight weeks. The scores did not differ
between the control limbs of the rhBMP-2 and a-BSM-treated dogs at four
weeks (power = 0.8 to detect a difference of 15%), but the scores for the
control limbs of the rhBMP-2-treated dogs were greater than those for the
control limbs of the a-BSM-treated dogs at eight weeks (p <
0.05).
The scores for retention of residual carrier were significantly lower at
eight weeks than at four weeks in each treatment group (p < 0.05). The
a-BSM-treated limbs had higher scores for residual carrier than did the
rhBMP-2-treated limbs at four and eight weeks (p < 0.05). The median score
for residual carrier in the rhBMP-2-treated limbs was 2.3 (range, 1.3 to 3.0)
at four weeks. At eight weeks, only one rhBMP-2-treated limb had evidence of
residual carrier (median score for the group, 0.3; range, 0 to 1.3). The
median score for residual carrier in the a-BSM-treated limbs was 3.7
(range, 2.5 to 4.0) at four weeks and 2.7 (range, 1.0 to 3.8) at eight weeks.
Interobserver precision errors were 6% for grading of union and 10% for
grading of residual carrier.
Force-Plate Analysis
There was a significant group effect (p = 0.01), time effect (p = 0.0001),
and group × time interaction (p = 0.0001) for time-integrated vertical
force (Fig. 4). Prior to the
surgery, the time-integrated vertical force was similar between the two
treatment groups and between the treated and control limbs in both groups. For
the rhBMP-2-treated limbs, the time-integrated vertical force was not
decreased, compared with the preoperative level, at four and eight weeks after
the surgery. For the a-BSM-treated limbs and the two untreated, control
groups, the time-integrated vertical force was less at four and eight weeks
than it was before the surgery (p < 0.05). The time-integrated vertical
force was greater for the rhBMP-2-treated limbs than it was in any other group
at four weeks (p = 0.05). There was no difference in the time-integrated
vertical force among the control limbs of the rhBMP-2-treated dogs, the
a-BSM-treated limbs, and the control limbs of the a-BSM-treated
dogs at four (p > 0.05, power = 0.8 to detect a difference of 28%) or eight
weeks (p > 0.05, power = 0.8 to detect a difference of 22%). At eight
weeks, the time-integrated vertical force for the rhBMP-2-treated limbs was
not significantly different from that for the contralateral, control limbs
(power = 0.8 to detect a difference of 18%), but it was greater than that for
the control and treated limbs of the a-BSM-treated dogs (p < 0.05).
There was no difference in the time-integrated vertical force between the
a-BSM-treated limbs and their contralateral controls at any time-point
(power = 0.8 to detect a difference of 20% preoperatively, power = 0.8 to
detect a difference of 27% at four weeks, and power = 0.8 to detect a
difference of 25% at eight weeks).
Mechanical Properties
There was no difference in nondestructive axial stiffness between the
rhBMP-2-treated limbs and their contralateral controls at eight weeks (power =
0.8 to detect a difference of 22%) (Table
IV). Mediolateral and anteroposterior bending stiffness were both
greater in the rhBMP-2-treated limbs than they were in their contralateral
controls or in the a-BSM-treated limbs (p < 0.05). The values for all
three nondestructive mechanical tests on both the rhBMP-2-treated limbs and
their contralateral controls were not significantly different from or greater
than the values for normal
tibiae32. The
results of the nondestructive mechanical tests did not differ between the
a-BSM-treated limbs and their contralateral controls (power = 0.8 to
detect a difference of 35% in axial compression, power = 0.8 to detect a
difference of 34% in mediolateral stiffness, power = 0.8 to detect a
difference of 38% in anteroposterior stiffness, power = 0.8 to detect a
difference of 32% in torsional stiffness, and power = 0.8 to detect a
difference of 26% in torsional strength). The measurements derived from all of
these mechanical tests in the a-BSM group were lower than the values for
normal tibiae32.
The bending stiffness of the untreated, control limbs of the
a-BSM-treated dogs was similar to the low end of the normal range. Both
bending stiffness values for the control limbs of the rhBMP-2-treated dogs
were greater than those for the control limbs of the a-BSM-treated
dogs.
Torsional stiffness and strength were both significantly greater in the
rhBMP-2-treated osteotomy segments than they were in their contralateral
controls (Table IV). The values
for the rhBMP-2-treated limbs were similar to the values obtained for normal
tibiae32. The
torsional stiffness of the control limbs of the rhBMP-2-treated dogs was
similar to the low end of the normal range, but the torsional strength was
lower than normal. The treated and control limbs of the a-BSM-treated
dogs had similar torsional stiffness and strength. The values for both
measurements were significantly lower than the range of values for normal
tibiae. The torsional stiffness of the control limbs of the rhBMP-2-treated
dogs was significantly greater than that of the control limbs of the
a-BSM treated dogs (p < 0.05). There was a strong trend for torsional
strength to also be greater in the control limbs of the rhBMP-2-treated dogs
than in the control limbs of the BSM-treated dogs (p = 0.057).
None of the rhBMP-2-treated osteotomy segments failed in torsion through
the original osteotomy defect, whereas all eight of the a-BSM-treated
limbs and eleven contralateral controls failed through the osteotomy site;
this was a significant difference between the a-BSM and rhBMP-2 groups
(p < 0.05).
Histologic Analysis
Histologic evaluation of calcified sections confirmed the radiographic
observations of osseous callus bridging the osteotomy site and filling the
osteotomy gap in all of the rhBMP-2-treated limbs. Incomplete bone-bridging
was observed at the same locations in the majority of the control limbs of the
rhBMP-2-treated dogs and the majority of the treated and control limbs of the
a-BSM-treated dogs (Figs.
5 and
6). Since all of the mechanical
failures of the rhBMP-2-treated segments occurred outside the osteotomy site,
in normal diaphyseal bone, histologic evaluation of the healing of the
osteotomy site could be performed in this group without interference by damage
due to mechanical testing. The presence of fracture lines from the mechanical
testing did not interfere with the histologic evaluation of healing in the
other groups. The mechanical testing procedure returned the specimens to their
original orientation, and the preexisting osteotomy site could be
distinguished easily from the mechanical testing artifact.
Histologic evaluation revealed that bridging callus was composed entirely
of bone in all of the rhBMP-2-treated limbs
(Fig. 5). In addition, the
osteotomy gap was completely filled with bone in seven of the eight limbs. In
the remaining limb, one cortex was bridged with bone and the opposite cortex
was bridged with a combination of bone and cartilage. In contrast, in the
majority of the control limbs of the rhBMP-2-treated dogs, the callus was
composed of a combination of cartilage and bone, cartilage alone, or a
combination of cartilage and fibrous tissue typical of ongoing stages of
endochondral callus formation (Figs.
5 and
6). Only three of the control
osteotomy sites in the rhBMP-2-treated dogs were bridged with bone across at
least one cortex. A similar composition of the callus and osteotomy gap tissue
was observed in the a-BSM-treated limbs and their contralateral
controls; two of the osteotomy sites were bridged with bone across at least
one cortex in these two groups. Tetracycline-labeling of new bone formation in
the callus and osteotomy gaps confirmed these observations (Figs.
5 and
6).
Only two of the eight rhBMP-2-treated osteotomy sites had histologic
evidence of residual a-BSM carrier at eight weeks after the surgery
(Fig. 7). The amount of
residual carrier was <5% of the total callus area. All eight
a-BSM-treated osteotomy sites had histologic evidence of residual
a-BSM carrier. In three of the eight sites, approximately 25% of the
total callus area was residual a-BSM. In the remaining five sites,
<10% of the total callus area was residual a-BSM. Although cellular
detail was difficult to ascertain in the 80 to 100-µm calcified sections,
there did not appear to be any substantial inflammatory response associated
with residual carrier in any of the groups. Ghosts of large cells within
resorption lacunae on the surface of the residual carrier suggested resorption
was occurring through a cell-mediated mechanism presumably involving
osteoclasts and giant cells (Fig.
7), as previously
described21,22,33.
This study demonstrated that a single percutaneous injection of rhBMP-2
combined with a rapidly resorbing calcium phosphate paste (a-BSM)
accelerates bone-healing compared with that seen with a-BSM alone or in
untreated, control osteotomy sites.
The design of this study was intended to mimic percutaneous injection into
a fracture site after injury. The injection was performed four hours after the
osteotomy to allow sufficient time for a hematoma to form in the gap. In
addition, we wished to allow the dog sufficient time to recover between the
surgery and the subsequent treatment while still completing the treatment
protocol on the same day as the surgery. In the clinical setting, injection
four hours after the injury probably would be considered early, as treatment
may occur from one to several days after the injury.
The scores for radiographic union, the callus area, and the histologic
findings all indicated that healing was more advanced in the rhBMP-2-treated
osteotomy sites than it was in the a-BSM-treated sites or the untreated,
control sites. In addition, there was evidence of remodeling of the newly
formed cortex in the rhBMP-2-treated osteotomy sites at eight weeks.
Weight-bearing profiles at four and eight weeks after the surgery indicated no
reduction in the time-integrated vertical force for the rhBMP-2-treated limbs
compared with the preoperative levels. In contrast, the time-integrated
vertical force at four and eight weeks was decreased, compared with the
preoperative levels, in the a-BSM-treated limbs and in both control
groups.
Mechanical testing demonstrated that the torsional stiffness and strength
of the rhBMP-2-treated tibiae were similar to those of intact canine tibiae
and greater than those of the a-BSM-treated limbs and the limbs in both
control groups at eight weeks after the surgery. Failure of all of the
rhBMP-2-treated tibiae outside the original osteotomy site during mechanical
testing confirmed that the healed rhBMP-2-treated tibiae were similar to
intact tibiae. In addition, the reduction in the amount of residual carrier in
the rhBMP-2-treated osteotomy sites compared with the amount in the
a-BSM-treated sites suggested accelerated resorption of a-BSM in
the presence of rhBMP-2.
Numerous studies of animals have demonstrated the capacity of BMPs in
combination with implantable delivery vehicles to accelerate the healing of
osteotomy and fracture
sites13,25,34,35.
In addition, BMPs have been used successfully to treat open tibial
fractures36 and
chronic tibial nonunions in human clinical
trials37. However,
little information is available on the use of BMPs administered with more
versatile, injectable delivery vehicles. A number of studies have demonstrated
the capacity of BMPs delivered in buffer solutions to accelerate both closed
fracture and osteotomy-site healing in animal
models14,38.
rhBMP-2 delivered in buffer accelerated closed fracture repair in
rats14 and repair
of 1.0-mm osteotomy defects in
rabbits38. However,
definitive evidence of accelerated bone-healing following treatment with BMP
delivered in buffer has not been demonstrated in larger animal models, to our
knowledge. Acceleration of healing with rhBMP-7 (OP-1) delivered in buffer was
reported in 1.5, 3.0, and 5.0-mm ulnar osteotomy defects in
dogs39 and in
closed tibial fractures distracted 5 mm in
goats13. In the
former study, the small number of animals in each treatment group was
insufficient for statistical analysis. In the latter study, torsional
stiffness but not torsional strength was greater at two weeks in the
rhBMP-7/buffer-treated animals than it was in controls. The torsional
stiffness and strength of the treated tibiae were only 22% and 10% of the
values for intact tibiae at this early stage of healing. However, at four
weeks, there was no difference in the biomechanical parameters between the two
groups. Torsional stiffness and strength in both the treated and the untreated
limbs were approximately 75% and 47% of the values for intact tibiae. No
difference in the radiographic or histologic rate of healing or in the
torsional biomechanical properties were found following percutaneous injection
of rhBMP-2 delivered in buffer in a nonhuman primate fibular osteotomy
model35. The lack
of efficacy in larger animal models with slower bone-healing times may be
related to the rapid clearance of rhBMP-2 from the treatment site following
delivery in buffer compared with that following delivery in an implantable
collagen sponge (4% retention compared with 40% retention at one week in a
rabbit osteotomy
model)34,38.
In the current study, both torsional stiffness and strength of the
rhBMP-2/a-BSM-treated tibial osteotomy sites were equivalent to those of
intact tibiae at eight weeks. A similar acceleration of osteotomy-site repair
has been demonstrated after use of the combination of rhBMP-2 and a-BSM
in a rabbit ulnar osteotomy
model24 and a
nonhuman primate fibular osteotomy
model19,40.
Calcium phosphate cements such as a-BSM have been used extensively in
both animal models and human clinical trials, mainly as bone-void fillers in
metaphyseal
locations41.
Several excellent review articles describe the chemical composition, physical
properties, biocompatibility, cell-mediated mode of resorption,
osteoconductive properties, and clinical efficacy of these
carriers20,41,42.
There are few published reports of the efficacy of calcium phosphate cements
in diaphyseal bone. Acceleration of cortical defect healing with a-BSM
has been demonstrated previously in rabbits and
dogs22. In
contrast, as was the case in the present study, rabbit and nonhuman primate
osteotomy
models24,40
did not demonstrate accelerated healing with a-BSM carrier alone
compared with healing in untreated control osteotomy sites.
In the present study, resorption of the carrier and induction of bone
formation was demonstrated on sequential radiographs. Previous studies have
demonstrated the rapid resorption of a-BSM by
osteoclasts21,22,33,43.
The addition of rhBMP-2 accelerated the resorption of the a-BSM carrier
in this study, as demonstrated by the reduced retention observed in the
rhBMP-2-treated limbs. Similarly rapid resorption of a-BSM in the
presence of rhBMP-2 compared with the resorption with a-BSM alone was
reported in other
studies24,40.
The results of this study clearly indicate that a single injection of
rhBMP-2/a-BSM significantly accelerates healing in a canine
mid-diaphyseal tibial osteotomy model compared with that in contralateral
control osteotomy sites or in sites treated with a-BSM alone. On the
basis of these results, it appears that the combination of rhBMP-2/a-BSM
may be a promising injectable treatment to accelerate healing of closed
fractures in humans.
Note: The authors acknowledge the assistance of the following
individuals during the completion of this research: Maria Aiolova, for the
formulation of a-BSM and pilot work performed in this study; Kei
Hayashi, DVM, PhD, for performing the mechanical testing; Ron McCabe, for the
initial instrumentation and mechanical testing; Vicki Kalscheur, HT, for the
histologic processing; and Yan Lu, MD, for the radiographic scoring.