The surgical procedure was performed in a 1-in (2.54-cm) critically
sized ovine femoral defect model. Following approval by the Institutional
Animal Care and Use Committee at our institution (where the procedure was
validated in two pilot sheep) and from the Animal Use and Care Commission of
the Canton of Grisons, Switzerland, the procedure was performed in thirty-five
skeletally mature female sheep, with a mean age of 3.96 years (range, three to
five years) and a mean weight of 59.7 kg (range 51 to 70 kg), by the same
surgical team (U.R.K. and T.F.R.) at the AO Research Institute in Davos,
Switzerland. General endotracheal anesthesia was used during the surgery with
concomitant spinal anesthesia to minimize the inhalational agent requirement
and provide early postoperative analgesia. All animals were pretreated with
intravenous cefazolin (1 g) prior to the initiation of the surgical
procedure.
Surgical Technique
After the sheep was anesthetized and positioned in the lateral position and
the limb was prepared for surgery with Betadine solution (povidone-iodine) and
sterile draping, the retrograde, custom-designed, 15-mm solid stainless-steel
interlocking femoral nail was inserted through a traditional medial
parapatellar approach. An entry point was made 1 cm anterior to the
intercondylar notch of the distal end of the femur, and the femur was
sequentially reamed up to a 16-mm diameter. Reamings from the distal part of
the femur were saved for later application to the donor segment docking site.
The lateral aspect of the femur was then exposed through a 15-cm incision and
blunt dissection through the intermuscular plane between the vastus lateralis
and the hamstring muscles. A 2.54-cm distal femoral defect was created with
use of an oscillating saw; care was taken to ensure that all of the periosteum
was removed to simulate a preexisting defect
(Fig. 2, A). Then, a
longitudinal incision was made in the periosteum, over the lateral femoral
cortex, after which the periosteum was elevated off of the underlying cortical
bone with use of a periosteal elevator. The dissection was performed
circumferentially around the femoral diaphysis, leaving an in situ,
vascularized periosteal flap with preserved soft-tissue and muscular
attachments. (Refer to the next section for a description of the differences
in treatment among the study groups.) Following the dissection of the
periosteum for 4.5 cm proximal to the defect, an osteotomy was made in the
femoral diaphysis 3.5 cm proximal to the diaphyseal defect. This osteotomy
created the 3.5-cm diaphyseal donor segment, which was eventually transported
distally into the 2.54-cm diaphyseal defect. The intramedullary nail was then
inserted retrograde through the distal end of the femur, the simulated defect,
the osteotomized donor diaphyseal segment, and into the proximal aspect of the
femur. The donor segment was then transported distally into the simulated
diaphyseal defect and was secured to the docking site by means of two number-5
Ticron sutures (Davis and Geck, Wayne, New Jersey) placed through drill-holes
and thus filling the defect zone at time zero. Prior to the insertion of the
proximal and distal interlocking bolts, a lamina spreader was used proximal to
the transported donor segment to distract the newly created donor defect site
to a length of 2.6 cm (Fig. 2,
B). Finally, the previously elevated musculoperiosteal
flap was closed with a running 2-0 Vicryl suture (polyglactin; Ethicon,
Johnson and Johnson, Somerville, New Jersey), creating a well-vascularized
tube, which circumferentially covered the proximal diaphyseal donor defect and
the exposed intramedullary nail (Fig. 2,
C). The previously mentioned distal femoral reamings were
packed around the donor segment docking site, and the wound was closed in
layers.
Postoperative Care
All animals were housed in the large animal facility at the AO Research
Institute during the postoperative period. Postoperative antibiotics (1 g of
cefazolin administered intravenously every eight hours for twenty-four hours)
and analgesia were administered during the early postoperative period. Animals
were given feed and water ad libitum. Weight-bearing was protected during the
initial six postoperative weeks with the use of a body sling, which permits
full weight-bearing but prevents excessive movement or walking and allows the
animal to rest all of its body weight in the sling for comfort. All slings
were discontinued six weeks postoperatively, and the sheep were permitted free
weight-bearing and walking in single animal stalls. Standard anteroposterior
and lateral radiographs of the involved femur were made at time zero and at
two, four, eight, twelve, and sixteen weeks postoperatively. All sheep were
killed with a lethal overdose of barbiturate at sixteen weeks postoperatively,
and the involved and contralateral, control femora were explanted for
high-resolution, micro-computed tomography on a Faxitron X-ray system
(Hewlett-Packard, McMinnville, Oregon) and for histological analysis.
Experimental Groups
In all groups, the aforementioned procedure was performed identically with
the exception of different treatments within and surrounding the donor site
defect zone, which defined the experimental groups
(Table I). Each group consisted
of seven sheep.
Group 1: In the control group, the periosteal flap was excised and
only the surrounding thigh musculature enveloped the diaphyseal defect. The
defect was expected to persist in this group because of the lack of osteogenic
periosteum.
Group 2: The periosteal flap was elevated exactly as described in
the surgical technique section. This group tested the ability of the in situ
periosteal flap alone to provide new bone formation within the diaphyseal
defect by providing osteoprogenitor cells, osteotropic factors from the
matrix, and intact vascularity.
Group 3: The periosteal flap was elevated exactly as described in
the surgical technique section. Morselized cancellous autograft was obtained
from the ipsilateral iliac crest and packed around the intramedullary nail
within the cortical defect, and the periosteal tube was closed around the
autograft. This group was designed to determine whether morselized bone graft
augments new bone formation derived from the periosteum
(Fig. 3, A).
Group 4: With use of a sharp chisel, the periosteal flap was
elevated with thin fragments of adherent cortical bone. This group was
designed to determine whether this in situ vascularized autograft augments new
bone formation derived from the periosteum
(Fig. 3, B).
Group 5: The periosteal flap was elevated with a chisel, as in
Group 4, leaving adherent vascularized chips of cortical bone. Additionally,
morselized autogenous cancellous bone graft obtained from the ipsilateral
iliac crest was packed around the intramedullary nail within the cortical
defect. This group combines the osteogenic and osteoinductive properties of
both in situ autograft and traditional autograft
(Figs. 3, A and
B).
Radiographic and Microcomputed Tomographic Analysis
Standard anteroposterior and lateral radiographs of the femora were
evaluated for bridging regenerate bone within the donor site defect zone,
healing at the donor segment docking site, and construct stability
(Fig. 4). Femoral blocks were
explanted in toto to include bone, callus, and the surrounding soft tissues,
and high-resolution radiographs (Faxitron) were made of each specimen. Each
specimen was then fixed and embedded in polymethylmethacrylate and was imaged
in a high-resolution computed tomography system (µCT 20; Scanco Medical,
Bassersdorf, Switzerland). Imaging and image-based measurements were carried
out by Scanco, on a contract basis and according to mutually defined protocols
designed to maximize accuracy and reproducibility of all measurements.
Measurements were fully blinded, as the engineer making the measurements had
no knowledge of specimen or group numbers. A 2-cm-high volume of interest was
centered (longitudinally) in the defect zone with the assistance of the
high-resolution (Faxitron) radiographs. Then 368 image "slices,"
comprising 74 µm each, were imaged through the volume (with a total
resulting height of 1.9832 cm). The radius of the cylindrical volume was
chosen so that it was constant across all specimens and included all
regenerated tissue. The contralateral, control femora were imaged identically,
with the height, radius, and relative location of the volume that was imaged
matched to that of the defect zone in the contralateral, experimental
femora.
Two standardized gray-level thresholds were set with use of a
hydroxyapatite calibration phantom to segment regenerated tissue into voxels
(0.074 mm × 0.074 mm × 0.074 mm), including predominantly soft
tissue, callus, or bone. The lower threshold was set to count voxels with a
mean hydroxyapatite density of between 600 and 800 mg/cm3. At a
density of <600 mg/cm3, voxels comprise only soft tissue.
Between 600 and 800 mg/cm3, voxels comprise callus at varying
stages of mineralization. At a density of >800 mg/cm3, voxels
comprise bone. For reference, measurements were made on bone removed to create
the defect zone at the time of the surgery; fully mineralized, skeletally
mature cortical ovine bone exhibited a mean density of >900
mg/cm3 (Scanco;
).
By counting the total number of voxels in each of the threshold ranges, it was
possible to compare the total volume of regenerated tissue comprising callus
with that comprising bone for each femur and to calculate a mean volume of
regenerate bone and callus, respectively, for each experimental and control
group. In addition, the mean density was calculated for each of these
respective volumes. Finally, Imax, the area moment of inertia, was
calculated to quantify the functional capacity of the regenerate to resist
bending loads.
Statistical Methods
Prior to the initiation of this project, a power analysis was performed to
determine the number of subjects for each group. Given the fact that the
present study was unprecedented in the literature, the magnitude of the
expected differences among the groups could not be precisely predicted prior
to carrying out the experiment. For a one-way analysis of variance analysis,
with the assumption that the mean difference among the group means is 15%
(d) and that the group means are distributed evenly, for a standard
deviation (s) of 8%, the effect size (d) is:
d=ds=158=1.875
The effect size f for a minimum dispersion of means (e.g., k = 5%)
is
f=d×12k=1.875×0.1=0.59
For an a = 0.05 and ß = 0.30 and considering the five groups, it
was determined that a sample size of seven animals per group would be
required.
To test the hypothesis and to determine the significance of intergroup
differences in the radiographic and microcomputed tomographic data,
analysis-of-variance testing was applied with the Fisher protected
least-significant-difference post hoc tests, and nonparametric testing was
carried out with use of Wilcoxon signed-rank tests (StatView; SAS Institute,
Cary, North Carolina). A p value of <0.05 was considered to be
significant.
Qualitative Evaluation of Radiographs
Qualitative evaluation of the postoperative radiographs revealed
that the defect zone persisted in all of the control animals in which the
periosteum was excised (Group 1). In two of these animals, a thin shell of
bone was visible in one cortex at sixteen weeks. Hence, in the absence of the
periosteal sleeve, the 2.54-cm defect persisted. In contrast, the defect zone
was bridged with regenerated bone in all twenty-eight animals that composed
the four study groups. Furthermore, the transport segment docking site healed
uniformly in all thirty-five sheep as evidenced by bridging callus on
radiographs (Fig. 4).
Microcomputed Tomographic Evaluation of the Volume, Density, and
Moment of Inertia of Regenerated Bone
Explanted femoral specimens were evaluated for mean volume and density of
the regenerate tissue comprising bone and callus with use of high-resolution
microcomputed tomography (Fig.
5). The volume of regenerate tissue consisting of callus or bone
was significantly greater for all groups retaining the periosteal sleeve
(Groups 2 through 5) than for the control group without the periosteal sleeve
(Group 1) (p < 0.02; Fig.
6). With the numbers studied, no other significant intergroup
differences in regenerate volume were observed. Furthermore, the area moment
of inertia was significantly higher in the groups that retained the periosteal
sleeve (Groups 2 through 5) than in the control group (Group 1) (p < 0.02;
Fig. 7), but no significant
intergroup differences were observed, with the numbers studied. Although the
group with vascularized cortical bone chips adherent to the periosteum but
without the addition of bone graft (Group 4) showed a mean maximal area moment
of inertia closest to that of the contralateral, control side, all treatment
groups showed a significantly lower area moment of inertia than the
contralateral, control side (p < 0.03;
Fig. 7).
Significant differences were observed among groups with regard to the mean
density of the bone and callus-tissue regenerate. The mean density of the
bone-tissue regenerate (Fig. 8)
was significantly higher in the group with vascularized cortical bone chips
adherent to the periosteum (Group 4) than in the groups with periosteum and
with or without bone graft (Groups 2 and 3; p < 0.02). When Groups 4 and 5
were compared, no significant difference in the density of the bone regenerate
could be seen with the addition of bone graft within the periosteum retaining
adherent vascularized bone chips on its inner surface (Groups 4 and 5);
however, there was a significant decrease in the density of the callus
regenerate (p < 0.05). The density of all regenerate bone, across
experimental groups, approached 90% of the density of the contralateral,
control side (Fig. 8).
No sheep were removed from the study because of a failure to thrive in
association with the surgery. Two sheep were removed from the study because of
complications not related to the surgery itself. One sheep from Group 2 was
killed on postoperative day 8 following an adverse event in which the sling
mechanism failed, resulting in fracture of the involved femur. This failure
was not attributed to surgical technique or the procedure itself; thus,
another animal was operated on and replaced the prematurely killed specimen
per protocol. Additionally, one sheep from Group 4 died unexpectedly from
respiratory failure at the six-week postoperative mark. Autopsy revealed a
lung mass and massive pleural effusion. Radiographs up to that point exhibited
satisfactory bone regeneration and stable hardware, and the sheep was
functionally well from a clinical, orthopaedic perspective. An additional
sheep had been added previously to this group because of concerns over a
nondisplaced proximal femoral fracture, which had occurred during insertion of
the locking bolt in one animal. This additional animal took the place of the
sheep that had died of respiratory failure. In the animal with the
intraoperative fracture, the fracture went on to heal without incident. Thus,
all five groups had seven sheep survive until the time that they were killed.
No wound infections or neurovascular complications were noted in any of the
animals. Three animals required a reoperation at the two-week postoperative
interval for simple percutaneous exchange of an unstable locking bolt,
yielding a reoperation rate of 9%. All three survived the sixteen-week
postoperative period without incident and without mechanical instability as
evidenced on sequential radiographs.
The new surgical procedure was shown to be effective in bridging a
critical-sized defect, in an ovine femoral model, in all experimental groups
in which the periosteal sleeve was retained. Without retention of the
periosteal sleeve, the defect persisted in all animals, resulting in nonunion
at sixteen weeks after the initial surgery. The basic concept of the procedure
is to transport healthy diaphyseal bone into a preexisting defect zone over an
intramedullary nail, immediately restoring the defect zone in one stage, while
creating a new defect surrounded by healthy periosteum. Sixteen weeks after
surgery, significant intergroup differences were observed in the density, but
not the volume, of regenerate tissue within the defect zone. Adherence of
vascularized cortical bone chips to the inner surface of the periosteum was
associated with a significant increase in the density of regenerate bone
(Groups 4 and 5); in these groups, the addition of bone graft (Group 5) was
associated with a significantly lower density of regenerate callus. The
maximum area moment of inertia, a surrogate for the capacity of the regenerate
tissue to resist bending loads, was not significantly different among the
treatment groups. Hence, for the purposes of this study, the additional step
involving bone-graft acquisition does not appear to confer improvements in
volume, quality, or mechanical function (in bending) of the regenerate tissue
within the defect.
One potential advantage of this new
technique22 is the
ease of its implementation compared with alternative treatments. The procedure
can be performed in a single stage by a surgeon with basic orthopaedic skills,
without the need for costly biologics, microvascular tissue transfers, or
lengthy treatment times in an external fixator. If it can be shown to be
effective in human patients, associated complications could be minimized as
there would be less dependence on patient compliance and intensive
postoperative monitoring and there would be a decreased need for second stage
bone-grafting to achieve union.
Limitations to the study are typical of those inherent in experimental
models. For example, the procedure was performed under idealized circumstances
in which a diaphyseal defect was simulated at the time of surgery. Thus, the
operative field was free of a preexisting injury zone, soft-tissue injury,
infection, and/or impaired vascularity, all of which can exist in the clinical
situation and compromise healing when dealing with defects arising from
trauma, infection, osteonecrosis, or tumor. Nonetheless, this procedure has
the potential to move the bone defect from a diseased area to a healthy
softtissue bed, which may provide the best possible basis for endogenous
repair by means of the
periosteum22,33,34.
Another limitation of the study relates to the choice of the animal model and
the inherent differences between human and sheep bone. Although such
differences exist, sheep bone is quite similar to human bone in geometric and
healing parameters as well as in periosteal
physiology22,33,34.
In order to make the model as physiologically relevant to the human condition
as possible, we used skeletally mature sheep, two to four years past the age
of physeal closure, to best approximate skeletally mature (young adult) human
bone.
In addition to the previously mentioned limitations, further studies on the
underlying biological mechanisms of bone regeneration and revascularization,
the transport segment, and repair at the docking site should be carried out
prior to clinical implementation. Although they extend beyond the scope of the
present study, parallel histological and computational modeling studies are
under way to elucidate these mechanisms.
Some of the most challenging defects in clinical practice exceed the defect
size used in this model. The defect size was chosen as a critical size from a
biological and biomechanical perspective, i.e., it would not heal or fail,
respectively, within the time-period of the study. That the 2.54-cm defect was
of a "critical size" was validated by the study, as it persisted
in the control animals over the course of the study. The sizing proved to be
prudent from a biomechanical perspective, as there were no hardware failures
over the course of the study. The osteoprogenitor population and vascularity
of the periosteum may be capable of regenerating bone in larger defects, but
further research is needed to confirm this, especially before applying this
method in humans.
Finally, the regenerate-intramedullary nail construct showed adequate
mechanical stability during the sixteen-week study period. After the initial
six weeks of protected weight-bearing, all of the study sheep walked freely
until they were killed. During this time, no fractures or construct failures
occurred, although three sheep had a reoperation to correct locking-bolt
instability. We chose not to conduct biomechanical tests after explantation of
the regenerate bone. Measurements of the maximal area moment of inertia in
bending were used as a surrogate for the capacity to resist bending loads, and
it was similarly decreased in all treatment groups compared with the
contralateral controls at sixteen weeks, although none of the treated femora
failed under physiologic loading in the protected vivarium environment.
In conclusion, using an ovine femoral critical-defect model, we proved the
efficacy of a recently developed surgical
technique23 for the
bridging of massive long-bone defects in a one-stage procedure exploiting the
intrinsic osteogenic potential of the periosteum while providing mechanical
stability through intramedullary nailing. In addition, there appears to be an
advantage to harvesting the transport bone in such a way that vascularized
bone chips adhere to the inner surface of the periosteal sleeve, as this
technique showed superior density in the regenerate bone compared with that in
other groups. ?