Direct attachment and subsequent healing of soft tissues such as tendon to
metallic devices would have broad application throughout the field of
orthopaedics. With the possible exception of cadaveric allograft prosthetic
composites1,2,
we are not aware of any reports on current implant designs that have shown
successful tendon and ligament biologic attachment sufficient to withstand
physiologic loading.
Recently, highly porous metal materials have shown promise for bone
ingrowth
applications3,4.
Porous metals, also known as metal foams, can be produced with interconnective
porosity coupled with a regular pore shape and size. One such material that is
currently commercially available for a variety of orthopaedic implants
involves the elemental metal tantalum fabricated with >80% interconnective
porosity with use of a metal vapor deposition technique (Trabecular Metal;
Zimmer, Warsaw,
Indiana)5,6.
This material has been utilized clinically for the fabrication of acetabular
components6, spine
fusion cages7, core
decompression
dowels8, and
uncemented patellar
implants9. While
bone ingrowth has been
documented3,7,10,11,
soft tissue has also occasionally been observed extending into this
material12-14.
The purpose of the current study was to design a new porous tantalum device
and to evaluate the extent and mechanical characteristics of tendon-healing in
an experimental canine model. We hypothesized that when the initial interface
mechanical environment is carefully controlled, this porous metal allows the
ingrowth of soft tissue with clinically relevant tendon-to-implant fixation
strength approaching that of intact tendon-to-bone insertion.
Device Design
Asoft-tissue-attachment implant, consisting of two custom-fabricated porous
tantalum washers that were fixed with a screw with no opportunity for tendon
reattachment or healing except to the prosthetic porous tantalum surface, was
devised for the present study (Fig.
1).
Experimental Design
The study included forty dogs (eighty tendons). The dogs were randomized
into four groups that were analyzed immediately postoperatively (seven dogs)
or at three weeks (eleven dogs), six weeks (eleven dogs), or twelve weeks
postoperatively (eleven dogs) (Excel; Microsoft, Redmond, Washington). In the
immediate postoperative group, all seven animals were evaluated
biomechanically and none were evaluated histologically. In the other three
groups, eight animals were randomized to biomechanical analysis and three were
randomized to histomorphologic evaluation.
Animal Model
The study protocol was approved by the Institutional Animal Care and Use
Committee and the Orthopaedic Department Research Committee and complied with
the procedures detailed in the Guide for the Care and Use of Laboratory
Animals, published by the National Institutes of
Health15.
A previously described tendon reattachment model was
used16. Surgical
experience was gained directly from the original authors at their home
institution. Forty purpose-bred hounds weighing an average (and standard
deviation) of 22.37 ± 1.04 kg were acquired from commercial vendors.
Skeletal maturity was confirmed on the basis of the date of
birth17, with all
dogs being more than one year old (range, 1.1 to 1.7 years old) and having
closed epiphyses at the time of surgery. Physical examination indicated that
the glenohumeral joints were without structural
abnormality18.
Surgical Technique and Postoperative Care
Dogs were fasted on the night prior to surgery. General anesthesia was
induced with ketamine (10 mg/kg) and diazepam (0.5 mg/kg) and was maintained
with 1% to 2% isoflurane, which was administered with a vaporizer. A mixture
of morphine and lidocaine was titrated intravenously for intraoperative pain
control.
At the time of surgery, the supraspinatus tendon was carefully dissected
from the glenohumeral joint capsule and the surrounding structures without
opening the joint. The tendon was detached sharply from the greater tuberosity
at the base of its insertion, and a 0-monofilament steel three-pass locking
whipstitch as described by Krackow et al. was placed in the most distal 0.5 cm
of the tendon to minimize repetitive traumatic manipulation of the tendon
during fixation to the
implant19. The
greater tuberosity was prepared by the removal of 3 mm of bone to allow for
the flush seating of the bottom washer. The supraspinatus tendon was examined,
and a Keith needle was used to mark the thickest portion of the tendon. A
2.5-mm drill was used to prepare the humerus for screw placement. Depth-gauge
measurement was obtained, and the proper screw length was determined as the
bone depth plus 6 mm for the two washers plus 3 mm for the tendon. The bottom
washer was placed into position. The tendon was positioned over the bottom
washer, and the top locking washer and screw were assembled. While tension was
maintained on the tendon with use of the locking stitch, an incision was made
in the tendon with a number-11 blade in line with the collagen fibers of the
tendon. A 3.5-mm self-tapping titanium screw with a taper-lock head was
engaged into the top washer, and the screw body was passed through the
thickest portion of the tendon, through the bottom washer, and into the bone.
A torque wrench was used to standardize the pressure between the two washers.
The tendon was sandwiched to 8 in-lb of torque and was allowed to creep. The
screw was retorqued to 8 in-lb after ninety seconds. A bone tunnel was made in
the proximal part of the humerus, 5 mm distal to the washer edge. The locking
whipstitch ends were hand-tensioned and secured to the bone with a friction
knot (Fig. 1). The wound was
closed in three layers. Immediate postoperative weight-bearing was
allowed20-22.
All animals were given intraoperative and four postoperative doses of
antibiotics (cephalexin, 30 mg/kg). Given the conflicting reports on the
influence of nonsteroidal anti-inflammatory drugs on
tendon-healing23,24,
narcotic postoperative analgesia (buprenorphine, 0.02 mg/kg) was chosen. Dogs
received preoperative and postoperative care from staff veterinarians and were
seen at least once daily by the operating surgeon (J.S.R.). At each
time-interval, dogs were killed by means of an intravenous overdose of
pentobarbital.
Clinical Function (Gait and Stance)
Load-bearing of the forelimb was measured with use of force-plate analysis
(custom force plate; NK Biotechnical, Minneapolis, Minnesota). Six separate
measurements of the maximum vertical ground-reaction force were recorded with
a turn key system on the day prior to surgery and immediately prior to the
time of death. Data were recorded at 100 Hz. At the time of testing, dogs were
allowed full access to the kennel run and were encouraged to walk freely over
the force-plate. The mean ground-reaction force was calculated on the basis of
the peak values for each gait cycle.
Mechanical Strength Evaluation
For specimens undergoing biomechanical testing, the implant and the humerus
were exposed. The supraspinatus muscle was dissected from the scapula, and the
distal part of the humerus was divided transversely at the supracondylar
region. Care was taken to remove the joint capsule. To eliminate the potential
for freeze-thaw effects on tendon strength and
stiffness25, this
muscle-tendon-implant-humerus construct was wrapped in surgical sponges,
soaked in 0.9 normal saline solution, and taken immediately for biomechanical
testing.
The humeral diaphysis was embedded in polymethylmethacrylate in an aluminum
mold (diameter, 4 in [10.2 cm]) and was secured in the testing machine. To
minimize muscle tearing as a mechanism of failure, a carbon dioxide gas
innercooled cryo-jaw was used to affix the supraspinatus muscle to a load cell
on a Universal Testing Machine (Model 858 Bionix; MTS Systems, Eden Prairie,
Minnesota). The muscle belly and myotendinous junction of each specimen were
kept frozen by the cryo-jaw to force failure at the tendon-implant interface
by freeze-fixing the weak myotendinous junction. Specimens were tested in
tension to failure at a displacement rate of 100 mm per
second26.
Force-deformation curves were recorded, and the strength and stiffness of the
construct were determined from these curves
(Fig. 2). The same technique
was used to test the contralateral muscle-tendon-bone construct.
Tendon Function: Muscle Volume
Following mechanical testing, the supraspinatus muscle was carefully
separated from the cryo-clamp and thawed in 0.9 normal saline solution to a
temperature of 38°C. Muscle volume was measured with use of the
water-displacement technique. The volume of the contralateral supraspinatus
muscle was measured with use of the same technique.
Histomorphologic Examination
For specimens undergoing histomorphologic evaluation, the supraspinatus
muscle was dissected from the scapula and the proximal part of the humerus was
transected with a saw. The supraspinatus tendon was sharply detached at the
myotendinous junction.
Undecalcified specimens were prepared for histological analysis. The
proximal part of the humerus, the implant, and the attached tendon were
immediately fixed in 70% ethanol and stored at —60°C. The specimens
were dehydrated in increasing concentrations of ethanol, defatted in acetone,
and embedded in Technovit Polymer (Heraeus Kutzen, Armonk, New York).
Longitudinal sections through the implants were cut to a thickness of between
140 and 160 µm with use of the Exakt saw, grinder, and polymerization
system (Exakt, Apparatenbau GmbH, Norderstedt, Germany). The specimens were
surface-stained with a trichromium stain (celestine blue and alum hematoxylin)
and van Gieson counterstain for collagen. The specimens were viewed under
transmitted light and under polarized light to identify any parallel collagen
fibers. To provide a quantitative measure of tissue ingrowth, the depth of
collagen-staining tissue was measured for the top and bottom washers in each
of the nine histological specimens. The minimum extent of tissue ingrowth in
the most central histological section of each sample was used to determine
tissue ingrowth for each specimen.
Statistical Methods
For the mechanical tests, a sample size of eight yields 80% statistical
power for the detection of a 1.2-standard deviation mean difference between
the experimental and control sides with use of a paired t test. A sample size
of eight also yields 80% statistical power to detect a 1.6-standard deviation
difference in means between the healing-time groups with use of two-sample t
tests.
Construct strength and stiffness changes over time were analyzed with use
of two-factor repeated-measures analysis of variance. Analysis of variance was
also used to determine the interactive effect of healing time on treatment
effect. The Ryan-Einot-Gabriel-Welsch multiple range test (alpha = 0.05) was
used as the analysis of variance test for strength and stiffness, with the
time-zero right leg data used as a control. The Pearson paired procedure for
simultaneous equality of means and variances was used to evaluate the axial
ground-reaction force gait differences between the preoperative and
postoperative analyses of each dog as well as for the volume differences
between the right and left supraspinatus muscles of each dog.
There were no immediate surgical complications, and all animals recovered
well in the postoperative period. There were no instances of wound breakdown
or infection. Full and fluid passive range of motion was found in all
shoulders immediately postoperatively and at the time of death. Activity and
walking were unrestricted throughout the
study19-21.
Clinical Outcome: Gait and Stance
Spontaneous weight-bearing and walking was noted immediately
postoperatively. Subjectively, dogs were seen to subtly favor the uninvolved
leg for the first few days after surgery as the pain medication was
weaned.
Force-plate ground-reaction force measurements demonstrated resumption of
preoperative function in the involved limb and correlated with the subjective
observation of normal gait and stance by three weeks. The initial average
ground-reaction force as percentage of body weight was 105.9% (95% confidence
interval, ±2.4%). No difference could be found between the initial
ground-reaction force and the force that the dogs placed on the involved limb
at three weeks (109.6% of body weight) (p = 0.6045). Interestingly, at six
weeks, the dogs placed significantly greater force axially through the
operatively treated limb (114.4% [95% confidence interval, ±8.1%] of
body weight) (p < 0.0339). However, no significant difference could be
demonstrated between the preoperative load-bearing force and the load-bearing
force at twelve weeks (102.3% of body weight) (p = 0.7985).
Mechanical Properties
Tendon-implant strength as a percentage of normal increased significantly
(p < 0.0014) throughout the experiment. The initial strength of the tendon
attachment device was 39% of that of the normal, intact, uninvolved
tendon-bone unit (95% confidence interval, ±17%)
(Fig. 3). Gross examination
showed that all seven involved specimens failed when the tendon pulled through
the attachment device. Each tendon-device construct failed as a unit, with the
tendon tearing across the interface with the metal washers.
At three weeks of in vivo healing, tendon-implant strength increased
relative to that seen in the control group. These implants failed at 67% of
the normal strength (95% confidence interval, ±16%). The mode of
failure had also changed by three weeks. While two of the reimplanted tendons
pulled through the implant in a manner similar to that in the initial fixation
group, six failed by incompletely delaminating from the implant surface.
Linear fibers of granulation-like tissue remained attached to the implant
after ultimate failure. Bone was also seen growing up through the bottom
washer, fixing the washer firmly to the proximal part of the humerus.
At six weeks, the tendon-implant strength was 99% (95% confidence interval,
±12%) of that of the control (normal) tendon. All eight supraspinatus
tendons failed at the myotendinous junction, leaving a cuff of tendon attached
to the implant.
At twelve weeks, tendon-implant strength was found to be 140% (95%
confidence interval, ±14%) of that of the normal tendon. In five cases,
failure occurred at the myotendinous junction as had been observed in the
six-week group. The remaining three failures occurred elsewhere in the
experimental system. One specimen failed by means of a transverse fracture at
the surgical neck of the humerus. Another failed by means of muscle
pull-through at the cryo-clamp. The third failed by means of a combination of
humeral fracture with partial pullout from the polymethylmethacrylate potting
jig.
The stiffness of the construct approached that of the normal supraspinatus
tendon (p < 0.0299) over the course of the study. The initial stiffness of
the implant-tendon device with no ingrowth was 47% (95% confidence interval,
±17%) of that of the normal supraspinatus tendon. After three weeks of
healing, stiffness had risen to 62% (95% confidence interval, ±17%) of
normal. By six weeks, stiffness had increased to 94% (95% confidence interval,
±11%) of normal. By twelve weeks, stiffness had increased to 130% (95%
confidence interval, ±18%) of normal prior to failure
(Fig. 3).
Tendon Function: Muscle Volume
The initial muscle volume in the involved shoulder was 97% of that in the
contralateral shoulder. This trend, while not significant (p = 0.1250), may
have been due to stripped tissue that was trapped within the implant's metal
pores at the time of strength-testing.
At three weeks, the supraspinatus muscle showed significant atrophy, with a
33% (95% confidence interval, ±4.6%) loss of volume (p < 0.01)
(Fig. 4). By six weeks, the
muscle had recovered to 81% (95% confidence interval, ±7.6%) of its
normal volume. By twelve weeks, muscle volume had increased to 92% (95%
confidence interval, ±7.1%) of its normal volume.
Histomorphology
Polarized microscopy qualitatively revealed collagen fibers attaching to
the porous metal surface. Collagen-specific van Gieson counterstained
microscopy qualitatively showed increased cellular density within the metal
trabeculae.
At three weeks, gross examination showed red granulation tissue. This
tissue became organized into radiating fibers when placed under tension.
Closer examination showed this tissue to have grown deeply into the bottom
washer; fixation to the top washer was less secure. Quantitatively, the depth
of tissue ingrowth was 0.43 mm (95% confidence interval, ±0.17) in the
top washer and 2.93 mm (95% confidence interval, ±0.06) in the bottom
washer at three weeks. At six weeks, histomorphologic examination showed that
this tissue had become more organized, especially at the periphery of the
implant. The depth of tissue ingrowth was 2.80 mm (95% confidence interval,
±0.11) in the top washer, and there was full ingrowth to 3 mm in the
bottom washer in all three specimens. By twelve weeks, gross examination
showed white tendinous tissue extending uninterrupted from the myotendinous
junction, entering the space between the washers, and filling the metal
trabeculae on both sides of the implant. There was collagen staining tissue
extending for 3 mm through both the top and bottom washers in all three
twelve-week specimens.
There was a stark contrast in tissue-metal interdigitation between the
porous tantalum metal interface and the smooth titanium surface of the top
washer (Fig. 5). There were no
tissue-titanium interactions seen in any of the specimens from any time-group
at the titanium cover or the screw interface. Fibers between the porous
tantalum washers appeared to be oriented in the direction of the tensile
forces, with fibers whirling around the central screw and becoming oriented
parallel to each other in line with the putative muscle force vector.
Organized tendinous tissue filled the metal voids in both the top and
bottom washers by twelve weeks (Figs.
6,
7, and
8). Sharpey-like fibers
appeared to insert onto the surface of the porous tantalum, streaming from the
metal trabeculae toward the long axis of the tendon
(Fig. 8).
To our knowledge, the present study is the first to show direct biological
tendon attachment to a prosthetic metal implant surface with near-normal
tendon-to-bone insertional strength and stiffness. Prior surgical
reconstructions that have utilized an intact tendon-bone complex have shown
favorable bone-to-bone
healing15,27-30.
Likewise, techniques whereby tendon can be securely fixed directly to bone
have proved successful for achieving a strong tendon-bone attachment in
selected
circumstances18,31-35.
When tendon is directly fixed to metallic implants, weak fixation and
mechanical failure under physiologic loading have been previously
observed36-40.
In a previous study, Hacking et
al.13 examined the
strength of subcutaneous paraspinal fibrous tissue attachment to blocks of
porous tantalum. They found that porous tantalum achieved three to five times
the pull-off strength of a beaded cobalt-chromium
surface13. There
have been anecdotal clinical reports of surgeons attempting to reattach tendon
(patellar
tendon41,42,
abductor complex10)
to porous tantalum implants.
In the present experimental study, a soft-tissue attachment device was
constructed to optimize the mechanical milieu thought to enhance
tendon-healing20,21,43-48.
This device consisted of two custom-made 3-mm-thick porous metal washers, a
central screw, and a metallic wire suture. Our results showed that this
construct achieved a stable implanttendon interface with increasing strength
over time, an indication of biologic healing.
To test the implant, we modified a canine supraspinatus tendon-healing
model developed and tested by Gottsauner-Wolf, Chao, and
associates16,25,38,39.
In those previous studies, an "enhanced tendon anchor" was
constructed of titanium mesh (pore size, 400 µm) and was used to reattach
tendon as well as tendon-bone blocks.
It is reassuring to note that our initial gait, strength, and stiffness
data agree with those previously reported in association with the enhanced
tendon anchor model. In a 2002 study, Inoue et al. found initial load-bearing
of 110.3% of total body
weight39. In the
present study, preoperative ground-reaction force was 105.9% of total body
weight.
Our strength and stiffness data also agree with published canine
supraspinatus tendon data. Inoue et al. found the average absolute strength
and stiffness of intact normal tendon to be 1098.5 N and 100 N/mm,
respectively39. The
average values for intact normal tendon strength and stiffness in the present
study were similar (912.30 N and 80.9 N/mm, respectively). Similarly, the
average values for initial pullout strength and stiffness in the present study
(290.1 N and 28.6 N/mm) were comparable with reported values for the spiked
washer (149 N and 26 N/mm) and the enhanced tendon anchor prosthesis (339 N
and 20 N/mm)25.
To our knowledge, no previous report or implant has shown comparable
evidence for direct tendon-to-metal healing over time. Inoue et al. reported a
decrease in functional weight-bearing on the involved limb from 110.3% to
82.8% at six weeks
postoperatively39,
whereas our study showed an increase in weight-bearing on the involved limb
from 105.9% to 114.4% at six weeks. Even after twelve weeks of healing and
after augmentation with bone graft, normal gait had yet to be achieved in
association with the enhanced tendon anchor prosthesis, with weight-bearing on
the involved limb averaging
94.5%39. In the
present study, dogs continued to exhibit a normal gait pattern and placed
102.3% of normal weight on the involved limb at twelve weeks. Moreover, when
used without bone graft augmentation, Gottsauner-Wolf et al. found that direct
tendon attachment to the enhanced tendon anchor prosthesis could achieve only
16% of the normal supraspinatus tendon insertion strength at sixteen weeks
following
surgery38. In the
current study, the porous tantalum attachment device showed 99% insertion
strength at six weeks postoperatively.
The twelve-week superphysiologic strength and stiffness results deserve
comment. While the standard deviations of these results did overlap the normal
range for our control tendons, several factors might have led to greater than
intact tendon strength and stiffness. First, the area available for tendon
ingrowth was effectively doubled in the implant used because the tendon was
sandwiched between two washers of porous metal. Second, in the present study,
the control, "normal" tendons were those in the uninvolved,
contralateral shoulders of the same animals that had initial fixation in the
involved shoulders. This decision was based on the findings of previous
studies involving the same model and also on the a priori conviction that once
surgery was performed on one shoulder, the other shoulder might not remain
"normal" because of the increased weight-bearing that might be
expected after surgery on the contralateral forelimb. However, when the
mechanical testing results were reassessed with use of each experimental
animal's time-matched contralateral shoulder as the control value, a similar
time-dependent increase in strength and stiffness was observed
(Fig. 9).
Demonstrating that these results are translatable to a variety of tendon
types will require additional studies that will consider anatomic location,
animal species, and type of tendon-muscle unit injury. Nevertheless, the
present study highlights the potential utility of porous tantalum as a
soft-tissue attachment and repair biomaterial. The ability to achieve
functional reattachment of tendon to a prosthetic surface should offer novel
opportunities for the development and modification of orthopaedic implants and
should help to improve patient outcomes in cases in which tendon attachment to
bone has been lost or compromised. ?