It is important to determine how much initial tension should exist in an
anterior cruciate ligament graft when the knee is unloaded because this
tension affects the surgical outcome of the reconstructed
knee1-9.
If the initial tension is too low, it adversely affects knee joint function as
a result of excessive anterior
laxity8. If the
initial tension is too high, it adversely affects knee joint function as a
result of posterior subluxation of the
tibia1,10-12
and inhibited knee
extension1. High
initial tension also has been linked to adverse consequences to the graft,
which include excessive graft wear at the femoral tunnel
margin4, poor
revascularization, myxoid degeneration, and inferior mechanical
properties5,9.
Therefore, it is important to identify factors that influence the initial
tension. By judiciously controlling these factors, initial tension can be
optimized so that knee stability is restored while the adverse consequences
associated with high initial tension are avoided.
One factor affecting the initial tension required to restore knee stability
to normal is the stiffness of the graft construct. At the time of surgery, the
stiffness of the construct is controlled primarily by the stiffness of
fixation because the fixation method is typically less stiff than the
graft13,14.
Depending on the stiffness of the fixation method, the stiffness of a
double-loop tendon graft-fixation complex can be varied tenfold at
implantation (i.e., from 24 to 259
N/mm)13,14.
After surgery, the stiffness of graft constructs has not been determined, to
our knowledge, but this stiffness is also expected to vary widely in the in
vivo environment15.
In either case, the amount of initial tension required to restore normal knee
stability with a high-stiffness construct should be less than that with a
low-stiffness construct. While this relationship between the stiffness of the
construct and the initial tension can be appreciated intuitively, we found no
studies that determined quantitatively the initial tension for a specific
graft-construct stiffness and the stiffness that minimizes the initial tension
while restoring the anterior load-displacement behavior to normal.
The first objective of our study was to determine the optimal initial
tension carried by a double-loop anterior cruciate ligament graft in full
extension such that motion of the tibia relative to the femur with 225 N of
anterior force applied to the tibia (i.e., a 225-N anterior limit of motion)
was normal at 30° of flexion for a range of graft-construct stiffnesses. A
related second objective was to determine how well the 225-N anterior limit of
motion at flexion angles other than 30° was maintained for different
graft-construct stiffnesses. While the stability of the reconstructed knee is
of primary interest as a dependent variable, the unloaded position of the knee
also is of interest. This is because knee joint function is affected adversely
as a result of posterior subluxation of the tibia as noted above. Thus, the
third objective was to determine how the unloaded position of the tibia was
affected by the graft-construct stiffness over flexion angles ranging from
0° to 90°.
Specimen Selection and Preparation
Ten fresh-frozen cadaveric knee specimens (average age of donors,
sixty-five years; range, thirty-seven to seventyfive years) were obtained from
tissue banks. The knee joints were evaluated radiographically and visually at
the time of anterior cruciate ligament reconstruction. Only specimens with no
evidence of degenerative arthritis were included in the study.
The specimens were prepared for experimentation by completely removing all
tissue 50 mm proximal to and 50 mm distal to the joint line down to the bone.
The bone was scraped to remove the periosteum. The fibula was fixed in its
relative position with a screw anchored in the tibia. The fibula was then cut
off approximately 70 mm distal to the joint line. After reaming the medullary
canals of the tibia and the femur until only cortical bone remained, steel
rods that were 10, 11, or 12 mm in diameter were fixed in the canals with
polymethylmethacrylate. The knee was wrapped in saline-solution-soaked gauze
to prevent desiccation of the remaining tissues.
Determination of Anterior and Posterior Limits of the Intact
Knee
Each knee was aligned, preconditioned, and tested in a load application
system16. The load
application system is a six-degree-of-freedom apparatus that can apply loads
to the knee in all degrees of freedom and measure the corresponding
displacements according to a joint coordinate
system17.
Flexion-extension is adjustable over the full physiologic range, and
unconstrained motion is allowed in the remaining degrees of freedom. For this
study, anterior-posterior force was applied and anterior-posterior
displacement was measured (resolution, ±0.1 mm). With use of the steel
rods to interface the specimen to the load application system, each specimen
was aligned with use of the functional axes method, which aligns the natural
axes of joint motion with those of the load application
system16. Once
aligned, the shafts of the tibia and the femur were potted in aluminum tubes
filled with polymethylmethacrylate, which were then clamped rigidly to the
load application system. The knee was preconditioned by applying a 50-N
stepwise load to 250 N in both the anterior and posterior directions to the
tibia for five cycles at 0° and 90° of flexion. This preconditioning
protocol produced a repeatable load-deflection
pattern16. Zero
degrees of flexion was defined as the position of the knee with an extension
moment of 2.5
N-m18.
Motion of the intact knee was measured at 0°, 30°, 60°, and
90° of flexion in random order. The tibia was loaded in 15-N steps to
incrementally increase the load from 0 to 45 N of anterior force, decrease the
load to 0 N, increase the load from 0 to 45 N of posterior force, decrease the
load to 0 N, and increase the load from 0 to 225 N of anterior
force19. With use
of a linearly variable differential transformer as a transducer to measure the
anterior-posterior displacement of the tibia with respect to the femur, the
0-N posterior limit of motion was defined as the position of the tibia at 0-N
force once the load on the tibia was decreased from 45 N of posterior force.
The 225-N anterior limit of motion was defined as the position of the tibia at
225 N of anterior force.
Reconstruction of the Anterior Cruciate Ligament
The knee was removed from the load application system, and the joint was
exposed with use of medial and lateral parapatellar incisions. The patella and
patellar tendon were reflected distally, the joint was inspected for
degenerative arthritis, and the anterior cruciate ligament was excised.
A double-loop graft was constructed from bovine extensor tendons with use
of the same technique for preparing a double-loop semitendinosus and gracilis
hamstring graft. A double-loop bovine tendon graft was used because it has
similar structural properties and is longer than a double-loop semitendinosus
and gracilis
graft20. The added
length of the double-loop bovine tendon graft ensured firm fixation in a
freeze clamp, which is not always possible with a loop of gracilis tendon. The
bovine tendons were harvested, muscle was removed, and excess tendon was
trimmed so that two tendons when folded in half side-by-side fit snugly inside
a 9-mm-diameter sizing cylinder (Sizing Sleeves; Arthrotek, Warsaw, Indiana).
The free ends of each tendon were trimmed to a slight taper and were sewn with
1-0 suture with use of a crisscross stitch to facilitate passage of the graft
in the
knee21,22.
To position the tibial tunnel, a 2.4-mm-diameter Kirschner wire was placed
with use of a tibial drill-guide (Howell Tibial Guide; Arthrotek). The tibial
drill-guide customized the placement of the Kirschner wire in the sagittal
plane by accounting for variations in knee extension and slope of the
intercondylar roof. This placement prevented the graft from impinging against
the intercondylar roof in full
extension23-26.
The placement of the Kirschner wire was assessed with use of
anterior-posterior and lateral radiographs of the knee in maximum extension
before drilling the tibial tunnel. In the coronal plane, the placement of the
Kirschner wire was considered acceptable when it met two radiographic
criteria: (1) the Kirschner wire entered the intercondylar notch between the
medial and lateral tibial eminences and (2) the Kirschner wire formed an angle
of 70° with the articular surface of the medial tibial plateau. In the
sagittal plane, the Kirschner wire was considered to have acceptable placement
when it lay 4 to 5 mm posterior and parallel to the slope of the intercondylar
roof 26. The tibial
tunnel was drilled over the Kirschner wire with use of a 9-mm-diameter
cannulated reamer.
The femoral tunnel was placed by inserting a 9-mm-diameter endoscopic
femoral aimer (Size-Specific Femoral Aimer; Arthrotek) into the intercondylar
notch through the tibial tunnel. The knee was flexed until the hook of the
aimer locked into place in the over-the-top position, and a 2.4-mm-diameter
Kirschner wire was drilled into the femur. A 30-mm closed-end femoral tunnel
was drilled with use of a 9-mm cannulated end reamer.
To allow the spring attached to the free end of the double-loop bovine
tendon graft to simulate the combined stiffness of a femoral and tibial
fixation method, the femoral fixation used in the specimen had to be much
stiffer than the stiffest spring (275 N/mm). Accordingly, a special procedure
was developed to create an ultra-high-stiffness femoral fixation. A 10-mm
lateral-medial tunnel was positioned 25 mm inside the femoral tunnel with use
of the U-Shaped Drill Guide (Arthrotek). A 4-mm-diameter steel rod was
centered in the lateral-medial tunnel with a plug inserted in the femoral
tunnel. The lateral-medial tunnel was then packed with polymethylmethacrylate
and forced into the cavities of the trabecular bone with use of threaded end
caps27. Once the
polymethylmethacrylate had hardened, the 4-mm-diameter rod was removed so that
the femoral tunnel plug could be extracted. The femoral tunnel was cleared of
polymethylmethacrylate with a curet. The rod was then reinserted in the cement
mantle to form the femoral fixation post. The stiffness of the steel
rod-cement-bone construct was conservatively estimated as 13,500 N/mm, and the
corresponding deflection under a 225-N anterior force was 0.02 mm.
Rationale for Determining the Stiffness of the Springs
To vary the stiffness of the graft construct, six springs (25, 75, 125,
175, 225, and 275 N/mm) were selected as representing the distribution of the
stiffness of different combinations of femoral and tibial fixation methods.
The overall stiffness of eighteen different combinations of femoral and tibial
fixation methods was calculated with use of available values for the stiffness
of each femoral and tibial fixation
method13,14
and a spring-in-series
analysis14. The
overall stiffness for these fixation combinations ranged from 18 to 269 N/mm
(Table I). The graft construct
stiffness corresponding to each spring was computed by including a 450-N/mm
graft20 in series
with the spring and ranged from 24 to 171 N/mm.
Determination of Anterior and Posterior Limits of the Reconstructed
Knee
A custom fixture that measured the graft tension and allowed the initial
graft tension to be adjusted and the effective stiffness of the fixation
method to be varied was added to the tibial unit of the load application
system (Fig. 1). Upon exiting
the tibial tunnel, the four limbs of the graft were gripped with a freeze
clamp28. A
load-cell (Futek Advanced Sensor Technology, Irvine, California) attached to
the freeze clamp measured the graft tension. To adjust the initial tension, a
threaded shaft and knurled end cap were attached to the load-cell. The shaft
passed through a spherical alignment bearing in a steel plate attached to the
tibial unit of the load application system. A coil spring was sandwiched
between the steel plate and the end cap that threaded onto the shaft so that
the spring was compressed when the graft was in tension. Turning the knurled
end cap allowed adjustment of the initial tension. When the end cap was
removed from the shaft, a coil spring of a different stiffness could be
installed. Because the tibia was clamped rigidly to the steel plate, which was
bolted to the load application system that allowed unconstrained motion in
five degrees of freedom, the initial tension created a compressive load
between the tibia and the femur and also caused posterior translation of the
tibia.
Following the reconstruction, the knee specimen was clamped in the load
application system in the identical position to that for the intact specimen.
After the graft was secured in the freeze clamp, an arbitrary initial tension
of >250 N was applied to the graft with the knee at 0° of flexion with
the stiffest spring installed. The reconstructed knee was subjected to the
same preconditioning protocol that was used for the intact knee.
After the graft was preconditioned, the initial tension was set to 100 N at
0° of flexion and the knee was passively flexed to 120° while the
graft tension was checked to ensure that it did not increase prematurely at an
early flexion angle. A premature increase in tension is an indication that the
graft is impinged by the intercondylar roof as a result of a femoral tunnel
being placed too
anteriorly29. If a
premature increase in tension was observed, then the specimen was removed from
the study. This check was performed for the 25, 175, and 275 N/mm springs.
The reconstructed knee specimen was subjected to the same
anterior-posterior-anterior loading cycle used for the intact knee. Once one
of six springs (25, 75, 125, 175, 225, and 275 N/mm) was randomly selected,
the initial tension was varied at 0° until the 225-N anterior limit
matched within 0.5 mm that of the intact knee at 30° of flexion. When the
loading cycle was completed, the knee was returned to 0°, the initial
tension was adjusted as required to maintain the initial tension to the value
that matched the 225-N anterior limit of the intact knee at 30° of
flexion, a flexion angle from the remaining three was randomly selected, and
the loading cycle was applied again. After all four flexion angles were
completed, another spring was randomly selected and the procedure was repeated
until measurements were obtained for all six springs.
Repeatability checks were performed systematically during the experiment to
ensure that carryover effects were not important. For every other spring, the
check included five anterior-posterior-anterior loading cycles at 0° with
the initial tension for the spring that restored the 225-N anterior limit to
that of the intact knee. Five cycles were chosen because the
anterior-posterior load-displacement behavior within a spring was repeatable
after that number. After all of the springs were tested, a repeatability check
was again performed for the first spring with its appropriate initial
tension.
Statistical Analysis
On the basis of a power law where the dependent variable was the average
initial graft tension, nonlinear regression analysis was used to determine the
optimal initial graft tension as a function of fixation stiffness and of
graft-construct stiffness. A two-factor repeated analysis of variance was used
to determine whether stiffness affected the difference in the 225-N anterior
limit of motion between the reconstructed and intact knees. The two factors
were the stiffness at six levels (25, 75, 125, 175, 225, and 275 N/mm) and the
flexion angle at three levels (0°, 60°, and 90°). The same
analysis of variance was used to determine whether stiffness affected the
difference in the 0-N posterior limit of motion except that the flexion angle
was at four levels (0°, 30°, 60°, and 90°). Because the first
analysis of variance did not reveal a significant interaction (p = 0.2768) and
the second analysis of variance revealed a significant (p < 0.0001) but
unimportant interaction, the differences in the 225-N anterior limit and the
differences in the 0-N posterior limit were averaged over all flexion angles
tested at each stiffness and a Tukey's test was performed comparing these
averages for all pairs of stiffness. Significance was set at p < 0.05.
As the fixation stiffness increased, the initial tension to restore the
225-N anterior limit to normal at 30° of flexion decreased so that the
initial tension was lowest for the highest stiffness fixation
(Fig. 2). The drop in the
initial tension was greatest when the fixation stiffness increased from 25 to
75 N/mm, and the drop steadily decreased as the fixation stiffness was
increased incrementally by 50 N/mm. Overall, the highest stiffness fixation
(275 N/mm) required an average of 73 N of initial tension, which was more than
three times less than the average 242 N of initial tension required by the
lowest stiffness fixation (25 N/mm). A power law was effective in describing
the relationship between the average initial tension and the fixation
stiffness and the average initial tension and graft-construct stiffness.
The fixation stiffness significantly affected the difference in the 225-N
anterior limit from that of the intact knee (p < 0.0001), and the highest
stiffness fixation restored the 225-N anterior limit closest to normal from
0° to 90° of flexion (Fig.
3). When averaged over the three flexion angles, the highest
stiffness fixation (275 N/mm) overconstrained the 225-N anterior limit by 1.0
mm, which was 3.6 mm less than the 4.6 mm of overconstraint of the 225-N
anterior limit with the lowest stiffness fixation (25 N/mm).
The fixation stiffness also significantly affected the difference in the
0-N posterior limit from that of the intact knee (p < 0.0001), and the
highest fixation stiffness restored the 0-N posterior limit closest to normal
from 0° to 90° of flexion (Fig.
4). The highest stiffness fixation (275 N/mm) overconstrained the
0-N posterior limit by 2.6 mm, which was 3.8 mm less than the 6.4 mm of
overconstraint of the 0-N posterior limit with the lowest stiffness fixation
(25 N/mm).
Because the initial tension is an important variable affecting surgical
outcome and because high initial tension has a number of adverse consequences,
the primary purpose of this study was to determine quantitatively how much
less initial tension was required to restore stability to normal at 30° of
flexion for a high-stiffness graft construct compared with a low-stiffness
graft construct. A secondary purpose was to determine how the graft construct
stiffness affected the anterior load-displacement behavior of the knee from
0° to 90° of flexion by examining the 225-N anterior limit and the 0-N
posterior limit. The key findings of this study were that a high-stiffness
construct required more than three times less initial tension than that of a
low-stiffness construct, while at the same time it best restored both the
225-N anterior limit and 0-N posterior limit to normal.
Methodological Issues
The method used to set the initial tension provided a reaction load on the
tibia, which may be an important practical consideration in anterior cruciate
ligament reconstructive surgery. For tibial fixation devices that require
wrapping of either a suture or a tendon around a post, the tension applied by
the surgeon manually is inherently reacted by the tibia. For tibial fixation
devices that do not require wrapping of the graft, such as double staples,
metal interference screws, and WasherLocs (Arthrotek), the manual application
of tension does not create a reaction load on the tibia, provided that the
tendons are not still attached to the tibia and that the manual pull is
reacted solely by foot-ground reaction loads. However, a reaction load on the
tibia can be created by means of fixtures manufactured and marketed
commercially for this purpose; examples include the Intrafix (Mitek, Norwood,
Massachusetts), the Graft Tensioner (Arthrotek), and the Tension Isometer
(MEDmetric, San Diego, California).
In the present study, even though the initial tension was developed through
the application of extra-articular tension, the extra-articular tension was
representative of the initial tension or intra-articular tension. As
demonstrated in an earlier
study22, friction
between either the graft bundles or between the graft bundles and the tibial
tunnel does not create a substantial reduction in the intra-articular
tension.
Although viscoelastic effects can change the initial tension in a graft
through load
relaxation20,30,31,
viscoelastic effects were eliminated as a confounding factor in the
experimental design. The initial tension was always maintained so that the
225-N anterior limit matched that of the intact knee at 30° of flexion.
Thus, the initial tension was the tension required to restore the 225-N
anterior limit to that of the intact knee in the absence of viscoelastic
effects. This was appropriate because the 225-N anterior limit of the intact
knee at 30° of flexion was determined in the absence of viscoelastic
effects. Recognizing that knee laxity in the clinical setting is usually
determined in the absence of viscoelastic effects, we designed the experiment
to mimic the clinical setting.
The stiffness of the graft construct was changed by varying the stiffness
of the fixation with use of a coil spring rather than an actual fixation
method. The advantage of the use of a coil spring to simulate the stiffness
was that the effects of this independent variable could be isolated for study.
In addition to providing stiffness, actual fixation methods also allow varying
degrees of lengthening in the region of fixation (e.g., slippage) particularly
under repeated
loading13,14,32.
Accordingly, use of actual fixation methods would have confounded the design
of our study because any change in the load-deflection behavior could have
been caused either by lengthening in the region of fixation or by the
stiffness.
The range of graft construct stiffnesses considered is believed to be
meaningful not only before biological incorporation of the graft in the bone
tunnels but also after, when the fixation stiffness and/or graft stiffness can
be expected to
change15. Because a
graft construct stiffness of <24 N/mm was believed to be unlikely following
biological incorporation, only construct stiffnesses greater than this value
were considered. Because the initial tension became increasingly insensitive
to the graft construct stiffness as the stiffness increased, it was not
necessary to test springs with stiffness higher than 275 N/mm.
Importance and Interpretation of Results
One important finding of the present study is that the fixation stiffness
(and hence the graft construct stiffness), when varied over a practical range,
profoundly affected the initial tension required to restore the 225-N anterior
limit of motion to normal (see Fig.
2). The importance of this finding is that the 169-N greater
initial tension for a low-stiffness construct compared with a high-stiffness
construct is of sufficient magnitude to adversely affect both knee joint
function and the graft.
One adverse effect upon knee joint function is an increased risk of an
unstable knee, particularly during the early healing period before biological
incorporation of the graft, because the fixation carries the majority of the
graft force. High initial tension causes high graft forces to occur during
both passive flexion as the knee is fully
extended1,18
and when the knee is
loaded1,3.
During cyclic loading, fixation devices exhibit viscoplastic behavior, which
is a permanent lengthening of the graft construct that increases with
time13,32.
Because the viscoplastic effect is often greater for lower-stiffness fixation
devices13 and
because graft forces are higher with these fixation devices, the risk for
permanent lengthening of the construct is increased.
Another adverse effect to knee joint function is that high initial tension
causes the tibia to subluxate posteriorly on the
femur1,10,12,
which in turn loads the posterior structures of the
knee1,12.
Posterior subluxation decreases the moment arm of the patellar tendon, thus
increasing the force that the quadriceps must produce to cause
extension10. Also,
a greater extensor moment is needed to reach full
extension1 because
graft tension increases as the knee is moved into full extension. These two
effects combine to increase the quadriceps force required to actively extend
the knee18.
An adverse effect to the graft is excessive wear at the entrance to the
femoral tunnel4. In
the study by Graf et
al.4, increasing the
initial tension by threefold decreased the fatigue life of patellar tendon
grafts almost fourfold. This relative increase in initial tension is
comparable with that determined in the present study for low-stiffness
compared with high-stiffness graft constructs. Although the graft that we used
was a different type from that used by Graf et al., the much higher initial
tension for a low-stiffness construct may decrease the fatigue life of
double-looped hamstring tendon grafts. Other adverse effects to the graft that
have been linked to high initial tension include poor revascularization and
myxoid degeneration that result in inferior mechanical
properties5,9.
The power law relation described in this study between the initial tension
and graft construct stiffness (Fig.
2) is important because it demonstrates that the initial tension
is not particularly sensitive to the construct stiffness as long as the
construct stiffness is greater than approximately 125 N/mm. However, if the
construct stiffness falls below approximately 100 N/mm, then the initial
tension starts to increase dramatically. Thus, the power law may be useful in
determining a practical lower limit on the construct stiffness where the
initial tension can still be maintained close to the minimum value for the
highest-stiffness graft construct.
Caution should be exercised in applying the power law to set the amount of
initial tension applied to a graft intraoperatively, for several reasons.
Additional factors affect the amount of tension carried by the graft
immediately postoperatively. These factors include seating of the fixation
device and settling in of the graft in the fixation
device33.
Viscoelastic effects such as load
relaxation20,30,31
can also affect the initial tension, though these effects will be recoverable
for the most part provided that the initial tension is minimal in the
mid-region of the flexion arc, from about 20° to
90°20. This
requirement will be satisfied by a high-stiffness fixation, which requires a
relatively small amount of initial tension. As a result of these additional
factors, the tension carried by the graft immediately postoperatively may not
be directly indicated by the extra-articular tension.
Another reason for caution is that even if the tension carried by the graft
immediately postoperatively is indicated by the extra-articular tension, the
tension carried by the graft postoperatively may not be maintained. This is
because lengthening of the graft construct can occur as a result of
lengthening in the region of the fixation due to repeated loading from
rehabilitative
exercises33 and/or
lengthening in the region between the fixations due to remodeling of the graft
in the in vivo
environment34,35.
The potential for lengthening in the region of fixation due to repeated
loading is reduced for high-stiffness constructs.
A final reason is that the stiffness of the graft construct can change in
the in vivo environment. After the graft has been incorporated into the bone
tunnels, the stiffness of fixation may either increase or decrease depending
on the type of fixation
device15. Thus, it
is possible that a low-stiffness construct at the time of implantation may
become a high-stiffness construct after biological incorporation of the graft.
Thus, it might be advantageous to achieve a low initial tension
intraoperatively rather than a high initial tension, assuming that the initial
tension is maintained.
Another important finding of this study is that both the 225-N anterior
limit and 0-N posterior limit were more closely restored to normal over the
flexion range from 0° to 90° for a high-stiffness construct (Figs.
3 and
4). On the average, the 225-N
anterior limit differed by 1.0 mm from normal while the 0-N posterior limit
differed by 2.6 mm from normal with both differences being
overconstrained.
Because both limits became more overconstrained as the stiffness decreased,
the difference in anterior laxity from that of the intact knee was relatively
constant for the various stiffnesses. Defining the anterior laxity as the
anterior displacement that the tibia undergoes from the 0-N posterior limit to
the 225-N anterior limit, the anterior laxity was 1.8 mm greater than that of
the intact knee at the lowest stiffness and 1.6 mm greater than that of the
intact knee at the highest stiffness. Therefore, even though a high stiffness
produced both a 225-N anterior limit and a 0-N posterior limit closer to
normal, this improvement in anterior load-displacement behavior, particularly
for the 0-N posterior limit, may not be evident clinically with a knee
arthrometer because this device measures anterior laxity and not the limits of
motion36,37.
In summary, the present study indicates that the initial tension is more
than three times less for a high-stiffness construct than for a low-stiffness
construct. Furthermore, both the anterior and posterior limits of motion are
closer to normal for a high-stiffness construct than the limits achieved with
a low-stiffness construct. When the initial tension is maintained, the
clinical relevance is that a high-stiffness graft construct likely avoids many
adverse consequences to both knee joint function and the graft that can result
from high initial tension. When the initial tension is not maintained, the
clinical relevance is that a high-stiffness construct likely will better
prevent a recurrence of knee instability.