Patients and Entry Criteria
From January 2000 to June 2001, 120 consecutive patients undergoing primary
anterior cruciate ligament reconstruction were randomized in a strictly
alternating manner to treatment with use of autogenous bone-patellar
tendon-bone or doubled hamstring tendon autografts. Patients were excluded if
they were adolescents with open physes or were more than forty years old; if
they had an acute lesion of the anterior cruciate ligament (i.e., the interval
between the injury and the operation was less than thirty days); if they had
other ligament tears or had undergone a previous operation in the same knee
(with the exception of a previous meniscectomy); if they had injured the
contralateral knee; if they had degenerative changes of the articular
cartilage (grade-III or IV changes according to the Outerbridge classification
system24) at
arthroscopy; if they complained of patellofemoral symptoms; and, if after
being informed about the purpose of the study before the operation, they
refused to reliably take part in the study up to the two-year follow-up.
Ethical approval was obtained from the internal review board. All subjects
were informed of the study procedure, the purpose of the study, and any known
risks, and all gave informed consent.
All 120 patients returned for four-month, one-year, and two-year follow-up
visits. The two groups were comparable with respect to age, sex, body weight,
height, generalized laxity, involved side, interval between the injury and the
surgery, preoperative activity level, and mechanism of injury. In both groups,
there were forty-six men and fourteen women, and there was an equal number
(thirty) of right and left knees. The mean age at the time of surgery was
twenty-five years (range, sixteen to thirty-nine years in the bone-patellar
tendon-bone group and fifteen to thirty-nine years in the hamstring group).
The average body weight and height were 74 kg (range, 49 to 102 kg) and 175 cm
(range, 155 to 201 cm), respectively, in the bone-patellar tendon-bone group
and 70 kg (range, 45 to 95 kg) and 174 cm (range, 156 to 195 cm),
respectively, in the hamstring group. On the basis of the Carter and Wilkinson
method25,
generalized joint laxity was recorded in thirteen patients in the
bone-patellar tendon-bone group and in eleven patients in the hamstring
group.
The majority of injuries were noncontact-type injuries that occurred during
sports activities, such as soccer (thirtyfour in the bone-patellar tendon-bone
group and thirty-five in the hamstring group) and skiing (five in the
bone-patellar tendon-bone group and six in the hamstring group), or during a
motor-vehicle accident (ten in the bone-patellar tendon-bone group and seven
in the hamstring group).
The mean interval between the injury and the operation was twenty-four
months (range, one to ninety-six months) in the bone-patellar tendon-bone
group and twenty-seven months (range, one to 123 months) in the hamstring
group. In the bone-patellar tendon-bone group, seven patients had surgery one
to three months after the injury, twelve patients had surgery three to six
months after the injury, and forty-one patients had surgery more than six
months after the injury. In the hamstring group, eight patients had surgery
one to three months after the injury, eight patients had surgery three to six
months after the injury, and forty-four patients had surgery more than six
months after the injury. There were nine competitive athletes in the
bone-patellar tendon-bone group and eleven in the hamstring group. The
remaining patients were recreational athletes. All but eight patients in the
bone-patellar tendon-bone group and seven patients in the hamstring group
played pivoting sports before the injury. Previous surgery included six
arthroscopic medial meniscectomies (10%) in each group.
Preoperatively, all knees had positive Lachman and pivot-shift tests. In
the bone-patellar tendon-bone group, the Lachman test was graded 1+ (3 to 5
mm) in nine knees (15%), 2+ (6 to 10 mm) in forty-seven knees (78%), and 3+
(>10 mm) in four knees (7%). The pivot shift was 1+ (glide) in eight knees
(13%), 2+ (clunk) in forty-eight knees (80%), and 3+ (gross) in four knees
(7%). In the hamstring group, the Lachman test was graded 1+ in seven knees
(12%), 2+ in forty-nine knees (82%), and 3+ in three knees (5%). The pivot
shift was 1+ in seven knees (12%), 2+ in fifty knees (83%), and 3+ in three
knees (5%). Anterior tibial translation was evaluated preoperatively with use
of the KT-1000 arthrometer, and the side-to-side difference at 30 lb (13.6 kg)
was an average of 7 mm (range, 4 to 12 mm) in both groups. Asymptomatic
patellofemoral crepitation was present preoperatively in nine knees in the
bone-patellar tendon-bone group and in eight knees in the hamstring group. No
significant differences were detected between the two groups with respect to
any of the above-mentioned categories.
Surgical Technique and Postoperative Rehabilitation Program
To standardize the surgical technique, we adopted the technique described
by Mariani et al.26
for the bone-patellar tendon-bone graft and by Howell and
Gottlieb27 for the
double-looped semitendinosus and gracilis tendon graft. The grafts and the
fixation devices used represent the only variations between the two
techniques. A one-year learning curve was considered to be sufficient to
become skilled in both methods and thus avoid any bias. All anterior cruciate
ligament reconstructions were performed by the senior author (P.A.) under
arthroscopic control and with the use of a tourniquet. The operation began
with evaluation of the intra-articular abnormality and treatment of the
meniscal lesions.
The bone-patellar tendon-bone graft was harvested through an 8-cm
longitudinal skin incision centered over the medial aspect of the patellar
tendon. After undermining the subcutaneous tissue, the paratenon was dissected
in order to define the tendon margins and was preserved for closure. The
central third of the patellar tendon, 9 to 11 mm in width, was removed with a
rectangular bone plug (20 to 25 mm in length) at each end. The tendon portion
of the graft was freed from fat, whereas the bone blocks were trimmed in order
to fit a 9, 10, or 11-mm-diameter bone tunnel. Five knees (8%) were treated
with a 9-mm graft; fifty-two (87%), with a 10-mm graft; and three (5%), with
an 11-mm graft. At the end of surgery, the paratenon was sutured to close the
defect while the bone defects were filled with autologous bone chips collected
during graft preparation and tunnel drilling.
The hamstring tendons were harvested through a 3 to 4-cm vertical skin
incision placed 2 cm medial to the tibial tubercle across the top of the pes
anserinus. Both tendons were delivered out of the wound with a curved clamp,
their distal expansion to the crural fascia was
severed28,29,
and the tendons were stripped to the proximal musculotendinous junction with a
smooth tendon stripper (Linvatec, Largo, Florida). The distal ends of the
tendons were left attached to bone and fascia. Retained muscle and fat tissue
were removed by blunt dissection with a periosteal elevator, and number-1
Vicryl absorbable sutures (polyglactin; Ethicon, Somerville, New Jersey) were
sewn to the tendon ends. In order to taper the tendons when tension was
applied to the sutures, one-quarter of the circumference of each tendon was
encircled with each throw of the suture to achieve a crisscrossing Chinese
fingertrap pattern. The midpoint of both tendons was then looped over a single
suture. This suture was used to pull the four-bundle graft through a series of
calibrated cylinders (Arthrotek, Warsaw, Indiana). The diameter of the
snuggest-fitting cylinder defined the diameter of the four-bundle graft and
the size of the cannulated reamer used to drill a snug bone tunnel. Seven
knees (12%) were treated with a 7-mm graft; forty-five knees (75%), with an
8-mm graft; and eight knees (13%), with a 9-mm graft. Once prepared, the graft
was rolled up and placed under the sartorius fascia to avoid contamination
before insertion.
To avoid graft impingement, a Kirschner wire was inserted into the tibia
with use of the One Step tibial guide (Arthrotek). With the knee in extension,
the intra-articular arm and the bullet tip of the guide were centered within
the intercondylar notch being constrained between the posterior cruciate
ligament, the lateral femoral condyle, and the roof of the notch. This
three-point fixation customized the orientation of the guide and allowed the
insertion of the guidewire without notch impingement. The extra-articular arm
of the guide has a hole for the insertion of a reference pin. Keeping this pin
parallel to the joint line (perpendicular to the tibial crest), the guidewire
was inserted in the frontal plane at an angle of 70° to the medial tibial
plateau30. The
tibial tunnel was then drilled with use of a cannulated reamer of the specific
graft size. During reaming, a cylindrical sleeve was pushed against the tibia
to collect bone debris, which later was used for grafting bone defects in the
bone-patellar tendon-bone group and bone tunnels in the hamstring group.
The need for a notchplasty was evaluated with use of an impingement rod of
the same diameter as the graft passed through the tibial tunnel into the notch
with the knee in full extension. If passage of the rod was obstructed, we
expanded the anterior notch using a motorized shaver until there was 2 to 3 mm
of clearance between the rod and the roof and the lateral wall of the notch,
allowing the rod to pass freely in and out of the notch.
The femoral guidewire was inserted with use of size-specific femoral aimers
through a transtibial approach, keeping the knee at about 80° of flexion.
The femoral guidewire was placed 5 mm anterior to the posterior cortex to
allow a 1 to 2-mm posterior cortical wall after reaming at about eleven
o'clock (right) or one o'clock (left). With use of a femoral aimer through a
transtibial approach, the entry site of the guidewire is constrained and there
is a consistent risk of being too vertical (the twelve o'clock position). To
avoid this, we aimed to the most inferior position on the wall of the notch;
varus, internal tibial rotation, and external rotation of the femoral aimer
helped in this step. The femoral socket was then reamed to the graft size with
use of a cannulated atraumatic reamer.
Fixation on the femoral side was transcondylar with use of a Tunneloc screw
(Arthrotek) in the bone-patellar tendon-bone group and a Bone Mulch screw
(Arthrotek) in the hamstring group. After proximal fixation of the graft, the
pistoning pattern of both types of graft was checked by applying tension to
the free ends for ten cycles of knee motion from full extension to full
flexion. In the hamstring group, traction on the two free graft ends also had
a further stabilizing effect on the joint. In fact, they acted as a cord
looped around a pulley (the nose of the Bone Mulch screw) so the pretensioning
forces were directly transmitted to the tibia through the other two graft ends
attached to the tibia compressing the tibia toward the femur. Tibial fixation
was achieved in extension with use of a soft threaded interference screw (Soft
Silck Cannulated Screw; Smith and Nephew Acufex, Mansfield, Massachusetts) for
the bone-patellar tendon-bone group and a WasherLoc device (Arthrotek) for the
hamstring group. A low manual tension of approximately 20 N was applied to
both grafts, to minimize the risk of a dangerous increase in graft tension
during full active knee extension.
In the hamstring group, at the end of surgery, the bone debris collected
during tunnel reaming was grafted in the femoral and tibial tunnels. With use
of specific compactors, the bone graft was compacted in the femur through the
body of the cannulated screw, whereas in the tibia it was placed directly from
the extra-articular end of the tunnel.
Postoperative anteroposterior and lateral radiographs (Figs.
1-A,
1-B,
1-C,
1-D) were made at the end of
each reconstruction to assess the correct placement and fixation of the
graft.
A brace-free, aggressive controlled rehabilitation protocol was adopted in
both groups. Passive range-of-motion exercises were instituted immediately.
The rehabilitation protocol was deemed to be aggressive (but not accelerated)
in order to try to restore full range of motion within the first month. A
written rehabilitation protocol with clear drawings and pictures of each
single exercise was also given to all of the patients in an attempt to achieve
maximum compliance. Knee swelling was managed with rest, ice, nonsteroidal
anti-inflammatory drugs, and partial weight-bearing. Muscle-strengthening
exercises were started on the first postoperative day with isometric
quadriceps contractions and progressed to active closed-chain exercises by
four to six weeks postoperatively. Patients were allowed full weight-bearing
three to five weeks postoperatively and returned to running at three months.
Return to sports-specific training was allowed at four months, and return to
competition was allowed at six months.
Follow-up Evaluations
All patients were evaluated before surgery, every two weeks up to the
second postoperative month, monthly up to four months after surgery, and then
at one and two years thereafter by an independent and blinded observer (F.G.).
At each evaluation, both knees of the patient were covered with a stockinette
in order to hide the involved side and the skin incision.
All patients were evaluated with use of a visual analog
scale31, the Knee
Injury and Osteoarthritis Outcome
Score32, the new
International Knee Documentation Committee (IKDC) evaluation
form33, and the
functional knee score for anterior knee
pain34.
At each follow-up visit, the range of motion of the involved knee in
relationship to that of the contralateral, normal knee was measured with use
of a long-arm goniometer. Extension deficit was determined with the subject
lying in the prone position and was measured as the difference in the heel
height of the involved limb in comparison with the passively fully extended
posture of the contralateral, uninjured limb.
A side-to-side difference in anterior tibial translation was assessed with
the knee flexed 30° with use of the KT-1000 arthrometer (MEDmetric, San
Diego, California) at 133 N and maximum manual
forces35.
Sensory changes possibly related to the surgical dissection of the
infrapatellar branch of the saphenous nerve were evaluated by asking the
patients to delineate the boundaries of the area using a dermographic pen. The
two major axes were measured, and the overall area was calculated in square
centimeters.
Concentric muscle strength recovery of extensors and flexors was measured
with use of an isokinetic dynamometer (Cybex NORM; Lumex, Ronkonkoma, New
York). Before testing bilateral seated knee extension and flexion, a
ten-minute warm-up was done on a stationary bicycle. Formal testing consisted
of ten maximal repetitions at 180°/sec, five maximal repetitions at
120°/sec, and five maximal repetitions at 60°/sec, testing the
uninvolved extremity first and the involved limb second. The arc of motion
recorded by the machine during the test ranged from full extension to 90°
of flexion. We also evaluated the muscle strength of the tibial rotators. In
this case, the patient was positioned as described by Hester and
Falkel36 and was
fitted with an appropriately sized Air-Stirrup ankle brace (Aircast, Summit,
New Jersey) to restrict ankle inversion and eversion. The foot was then
affixed to a foot-plate at 30° of dorsiflexion with use of two crossing
Velcro straps and a heel post to restrict foot motion. Five warm-up
repetitions were performed before each measurement. Formal testing consisted
of ten maximal repetitions at 90°/sec, five maximal repetitions at
60°/sec, and five maximal repetitions at 30°/sec, testing the
uninvolved extremity first and the involved limb second. Mean peak torque
(measured in Nm) was established by averaging the five maximal effort
repetitions for each tested velocity. The strength deficits of the involved
limb were calculated by subtracting the mean peak torque of the involved knee
from that of the normal knee. The result was then divided by the mean peak
torque of the normal knee and was expressed as a percentage.
An anteroposterior weight-bearing radiograph, lateral radiograph with the
knee in full passive extension, and posteroanterior tunnel radiograph were
made at four months, one year, and two years postoperatively. All images were
centered with an image amplifier. In the sagittal plane, we measured the
position of the anterior aspects of the femoral and the tibial tunnel using a
previously described
method37.
Radiographic evidence of graft impingement was investigated with use of the
method described by Howell and
Clark38. On the
tunnel radiograph, we measured the angle between the central axis of the
tibial tunnel and a line tangent to the tibial plateau
(Fig. 2). On the basis of the
studies by Howell et
al.30, it has been
demonstrated that the correct tibial tunnel position has an angle of
<70°39,40.
Tunnel enlargement was measured, according to the method described by
L'Insalata et
al.41, by two
independent observers (F.B. and F.S.) on the anteroposterior and lateral
radiographs. With use of a caliper, the distance between the sclerotic margins
of each tunnel was measured at its widest dimension, together with the
diameter of the femoral fixation device. (The size of the drill-bit used for
tunnel reaming was recorded at the time of surgery, and the diameter of the
femoral fixation device was known and constant.) All measurements were
corrected for magnification, and change in the tunnel size was calculated as a
percentage of the diameter of the drill-bit. On the lateral radiograph, the
prevalence of femoral tunnel widening was not assessed because of the
superimposition of the fixation device on the femoral tunnel margins.
Statistical Methods
Before the investigation was initiated, the sample size was estimated on
the basis of the hypothesis that there was no difference in anterior-posterior
knee laxity between the treatment groups. A clinically relevant difference
between the groups was considered to be a 1-mm increase in anterior knee
laxity compared with the contralateral side. The standard deviation, as has
been seen in a previous
trial42, was set at
1.5 mm. A power calculation was performed with a confidence level of 95%
(a = 0.05) and a power (1-ß) of 90%. This yielded an estimated
sample size of forty-eight patients per group, and, when combined with an
expected rate of patients lost to follow-up of 20% at two years, a sample size
of sixty patients per group, or a total of 120 patients, was required. All
statistical analyses were conducted on Statistica for Windows (5.1 edition;
StatSoft, Tulsa, Oklahoma). A comparison of the differences between the groups
was made with the Student t test for continuous variables and with the
chi-square test and Fisher exact test for categorical variables. In all tests,
an alpha level of 0.05 was considered significant.
Meniscal Lesions and Treatment
The medial meniscus was torn in twenty-nine knees (48%) in the
bone-patellar tendon-bone group and in twenty-six (43%) in the double-looped
semitendinosus and gracilis tendon group. The lateral meniscus was torn in
eight knees (13%) in the bone-patellar tendon-bone group and in seven (12%) in
the hamstring group. Two meniscal lesions (medial and lateral) were recorded
in eight (7%; five in the bone-patellar tendon-bone group and three in the
hamstring group) of the 120 knees. A partial medial meniscectomy was performed
in thirty-seven knees (31%; nineteen in the bone-patellar tendon-bone group
and eighteen in the hamstring group). A partial lateral meniscectomy was
performed in fifteen knees (12.5%; eight in the bone-patellar tendon-bone
group and seven in the hamstring group). Nine knees (7.5%; four in the
bone-patellar tendon-bone group and five in the hamstring group) with a
reparable (longitudinal in the red-red zone of the posterior horn) medial
meniscal tear were repaired with an inside-out
technique43.
Meniscal repair sutures were tied after the anterior cruciate ligament
reconstruction was completed. A stable longitudinal lesion (<1 cm) of the
medial meniscus was left untreated in six knees (10%) in the bone-patellar
tendon-bone group and in three knees (5%) in the hamstring group.
Complications and Additional Surgery
No intraoperative or postoperative complications occurred in this series.
No patient underwent additional surgery of the knee.
Subjective Functional Assessment
The mean visual analog scale score, Knee Injury and Osteoarthritis Outcome
Score, and IKDC score showed a progressive and substantial increase during the
review period compared with the preoperative condition
(Table I). All patients were
satisfied with the postoperative result of the reconstruction. No significant
differences were found between the two groups with respect to the subjective
assessments at each follow-up evaluation. Furthermore, no correlation was
found between subjective assessment and the postoperative knee stability,
range of motion, muscle strength recovery, sports activity level, or
radiographic evaluations.
Range of Motion
Extension deficit was assessed with the patient lying in the prone position
and was measured as the difference in the heel height of the involved limb in
comparison with that of the contralateral, uninjured limb in full passive
extension. At the four-month follow-up evaluation, only three knees in the
bone-patellar tendon-bone group showed a 3° to 5° extension loss. No
patient in the hamstring group showed an extension deficit. At the two-year
follow-up examination, there was a progressive recovery of knee extension and
only one knee in the bone-patellar tendon-bone group had a persistent 3°
extension loss.
Full flexion, or a loss of <6° of flexion in comparison with the
contralateral knee, according to the IKDC definition, was found in all knees
in the bone-patellar tendon-bone group and in fifty-nine knees in the
hamstring group at two years. One knee in the hamstring group had a 10°
loss of flexion. There were no changes with respect to the recovery of knee
flexion between four months and two years. No significant differences were
found between the two groups. No correlation was found between postoperative
recovery of the range of motion and knee stability, patellofemoral crepitus,
sports activity level, muscle strength recovery, or radiographic
evaluation.
Clinical Ligament Evaluation
Anterior tibial translation as demonstrated by the Lachman test was
restored to within 5 mm (1+), and with a firm end point, in all patients for
all follow-up visits for up to two years after surgery. At the two-year
follow-up examination, a pivot-shift glide (1+) was recorded in ten knees
(17%) in the bone-patellar tendon-bone group and in eleven knees (18%) in the
hamstring group. No patient reported symptoms of giving-way or showed a
pivot-shift clunk (2+). In the hamstring group, a correlation was found
between medial meniscectomy and an increased prevalence of a pivot-shift glide
at the time of follow-up. Eight of the twenty-four patients who had undergone
a medial meniscectomy previously or at the time of the anterior cruciate
ligament reconstruction showed a pivot-shift glide, whereas only four of the
thirty-six patients who had not had a medial meniscectomy showed a pivot-shift
glide (p = 0.035). An analogous correlation was not found in the bone-patellar
tendon-bone group. No correlation was found between the postoperative knee
ligament evaluation and the range of motion, muscle strength recovery, sports
activity level, patellofemoral crepitus, or radiographic evaluation.
Instrumented Testing
At the two-year follow-up examination, the average KT-1000 arthrometer
values for side-to-side differences were 1.95 mm (range, —1 to 5 mm) in
the bone-patellar tendon-bone group and 2.2 mm (range, 0 to 5 mm) in the
hamstring group when tested at 134 N. The 134-N and maximum manual
side-to-side KT-1000 results were comparable, with =2 mm in thirty-nine
patients (65%) in the bone-patellar tendon-bone group and thirty-four (57%) in
the hamstring group and between 3 and 5 mm in twenty-one patients (35%) in the
bone-patellar tendon-bone group and twenty-six patients (43%) in the hamstring
group.
The KT-1000 side-to-side anterior tibial translation both at 134 N and at
maximum manual force decreased significantly between the preoperative
examination and the two-year follow-up examination for both groups (p <
0.001). No significant differences were found between the two groups in terms
of the two-year postoperative arthrometric values. There were also no
significant differences in the clinical data and instrumented testing with
respect to gender, height, body weight, generalized laxity, or
meniscectomy.
Patellofemoral Symptoms
At the two-year follow-up evaluation, moderate, but asymptomatic,
patellofemoral crepitation was recorded in thirteen knees (22%) in the
bone-patellar tendon-bone group and in fourteen knees (23%) in the hamstring
group. The average functional knee score for anterior knee pain was 47 points
(range, 35 to 50 points) in the bone-patellar tendon-bone group and 48 points
(range, 34 to 50 points) in the hamstring group. No significant difference was
found between the two groups with respect to preoperative and postoperative
patellofemoral symptoms. No correlation was found between postoperative
extensor strength recovery and the presence of patellofemoral crepitus.
Harvest Site Abnormality
At the last follow-up examination, we found that a greater number of
patients complained of kneeling discomfort in the bone-patellar tendon-bone
group than in the hamstring group. Thirty-seven patients (62%) in the
bone-patellar tendon-bone group had kneeling discomfort compared with nine
(15%) in the hamstring group (p < 0.01). With regard to the infrapatellar
branches of the saphenous nerve, forty-six (77%) of the sixty patients in the
bone-patellar tendon-bone group and thirty (50%) of the sixty patients in the
hamstring group complained of alteration in anterior knee sensitivity. The
average area of the skin sensitivity disturbance was 40 cm2 (range,
10 to 88 cm2) in the bone-patellar tendon-bone group and 25
cm2 (range, 10 to 80 cm2) in the hamstring group; the
difference was significant (p < 0.001).
Overall IKDC Score
At the two-year follow-up examination, thirty-eight knees (63%) in the
bone-patellar tendon-bone group and thirty-four knees (57%) in the hamstring
group were graded A. The remaining knees were graded B, and no knees were
classified as C or D. No significant difference was found between the
groups.
Muscle Strength Recovery
The extensor and flexor muscles showed progressive recovery with time, and
the strength was comparable with the contralateral side in both groups at the
time of the two-year follow-up examination
(Table II).
Extensor muscle strength improved over time in both groups and became
comparable with that on the normal side at two years. At all angular
velocities in both groups, the improvement in strength between the four-month
and the one-year tests was significant (p < 0.005 for the bone-patellar
tendon-bone group, and p < 0.04 for the hamstring group). In the
bone-patellar tendon-bone group, the strength improvement between the one-year
and the two-year test was also significant (p < 0.005). With regard to
flexor strength, a significant difference in terms of better performance was
recorded only in the bone-patellar tendon-bone group between the four-month
and one-year follow-up examinations (p = 0.02). However, no significant
difference was found between the two groups of patients with respect to
postoperative flexor and extensor muscle strength status at each follow-up
evaluation.
Internal and external rotation strength recovery showed a similar
progressive recovery with time and became comparable with that of the
contralateral side in both groups by the two year follow-up evaluation
(Table III). At the last
follow-up evaluation, the internal rotation deficit at the three angular
velocities ranged from 1% to 2% in the bone-patellar tendon-bone group and 6%
to 7% in the hamstring group. However, no significant difference was found
between the two groups of patients with respect to the postoperative rotator
muscle strength.
Postoperatively, no correlation was found between residual muscle strength
deficit and knee ligament stability, sports activity level, patellofemoral
crepitation, or subjective functional assessment.
Activity
The IKDC form is used to grade the level of activity, with level I
indicating strenuous activity (cutting, jumping, and twisting); level II,
moderate activity (skiing, playing tennis, and performing heavy manual labor);
level III, light activity; and level IV, sedentary activity. Before the
rupture of the anterior cruciate ligament, fifty-two (87%) of the sixty
patients in the bone-patellar tendon-bone group and fifty-three (88%) of the
sixty patients in the hamstring group were involved in level-I or II
activities. At the two-year follow-up evaluation, twenty-seven patients (45%)
in the bone-patellar tendon-bone group were involved in level-I activities;
seven (12%), in level-II activities; twenty-two (37%), in level-III
activities; and four (7%), in level-IV activities. In the hamstring group at
the time of the two-year follow-up, twenty-nine patients (48%) were involved
in level-I activities; thirteen (22%), in level-II activities; eleven (18%),
in level-III activities; and seven (12%), in level-IV activities. A
significant difference was found with regard to the number of patients active
in level-I or II sports before the anterior cruciate ligament tear and after
the reconstruction. Postoperatively, significantly fewer subjects in both
groups were able to return to higher-level sport activities (p = 0.003 for the
bone-patellar tendon-bone group and p < 0.001 for the hamstring group).
With respect to patient-specific changes in activity level from before the
injury to after the surgery, thirty-four (65%) of fifty-two patients in the
bone-patellar tendon-bone group and forty-two (79%) of fifty-three in the
hamstring group returned to level-I or II sport activities. Only five patients
(10%) in the bone-patellar tendon-bone group and six (11%) in the hamstring
group reported that the decrease in sports activity level was related to knee
symptoms. No patient in either group was unable to return to sports because of
knee symptoms. Postoperatively, no significant difference was found between
the two groups with respect to the number of patients active in sports or the
level of participation.
Radiographic Assessment
The anterior margin of the intra-articular exit of the tibial tunnel in the
sagittal plane was located, on the average, at 42% (range, 32% to 52%) of the
width of the tibial plateau in the bone-patellar tendon-bone group and 40%
(range, 24% to 50%) of the width of the tibial plateau in the hamstring group.
On the basis of the radiographic measurement method of Howell and
Clark38, moderate
graft impingement was present in seventeen knees in the bone-patellar
tendon-bone group and in eighteen knees in the hamstring group. No case of
severe graft impingement was noted. At the time of follow-up, no correlation
was found between the presence of moderate impingement and increased anterior
tibial translation. The anterior margin of the femoral tunnel was located, on
the average, at 67% (range, 63% to 71%) of the femoral condyle width for the
bone-patellar tendon-bone group and at 63% (range, 55% to 69%) for the
hamstring group on the lateral projection radiograph. No correlation was found
between the position of the femoral tunnel and knee stability.
Tibial tunnel angulation with respect to the medial tibial plateau in the
frontal plane was an average of 66° (range, 60° to 78°) in the
bone-patellar tendon-bone group and an average of 69° (range, 60° to
76°) in the hamstring group; the difference was not significant. No
correlation was found between tibial tunnel angulation and increased knee
laxity or loss of knee flexion.
The amount of tibial tunnel widening in the sagittal plane was 55% in the
bone-patellar tendon-bone group and 60% in the hamstring group. In both
groups, the average tibial tunnel widening was 25% (range, 20% to 50%). In the
coronal plane, tibial tunnel widening was found in 32% of the knees in the
bone-patellar tendon-bone group and in 39% of those in the hamstring group.
The average widening of the tibial tunnel was 24% (range, 20% to 50%) in both
groups. Femoral tunnel widening in the coronal plane was observed in 17% of
the patients in the bone-patellar tendon-bone group and in 51% of the patients
in the hamstring group. This rate of femoral widening was significantly
greater in the hamstring group (p < 0.01). The average amount of tunnel
widening in the coronal plane in the femur was 23% (range, 20% to 30%) in the
bone-patellar tendon-bone group and 27% (range, 20% to 50%) in the hamstring
group.
The frequency and the amount of tunnel widening showed no changes after the
one-year follow-up evaluation. No correlation was found between tunnel
widening and knee stability.
The popularity of the use of hamstring tendons in anterior cruciate
ligament reconstruction has increased in recent years. Compared with the
bone-patellar tendon-bone graft, however, the initial results, in terms of
stability and clinical results, were
inferior2,44,45.
Recent investigations have found superior material properties of the equally
tensioned double-looped semitendinosus and gracilis tendons
graft46 compared
with the bone-patellar tendon-bone graft. Furthermore, the mechanical
properties of hamstring tendons seem to be preserved with increasing age, in
contrast to the bone-patellar tendon-bone graft, which seems to weaken with
age47.
The initially inferior clinical results with the hamstring graft could be
explained by inadequate graft fixation. Steiner et
al.48 were the
first to demonstrate that a direct and strong fixation of the hamstring graft
to bone was the key to success. New fixation devices have been introduced to
improve fixation of the hamstring graft, and the clinical results have
improved in terms of patient satisfaction, joint stability, and sports
activity
recovery49,50.
The present prospective, randomized clinical trial was performed to compare
double-looped semitendinosus and gracilis tendon and bone-patellar tendon-bone
grafts with use of newer surgical techniques and fixation devices. The
fixation devices for the hamstring graft were chosen on the basis of their
excellent biomechanical
properties20-23.
The femoral fixation in the bone-patellar tendon-bone group was selected to
resemble the transcondylar fixation used for the hamstring graft. At the
two-year follow-up evaluation, no significant difference was found between the
groups with respect to the IKDC scores. Kneeling discomfort was more frequent
in the bone-patellar tendon-bone group (p < 0.01), and femoral tunnel
widening was more frequent in the hamstring group (p < 0.01). All patients
were satisfied with the outcome of the operation, and three different
subjective scores failed to reveal any significant difference between the
groups. The knee range of motion was recovered in 100% of the patients in both
groups.
Knee stability was comparable in the two groups, with the exception of a
correlation that was found between the presence of a pivot-shift glide and
medial meniscectomy in the hamstring group (p = 0.035).
In contrast to our series that was reported in
19941, in which the
recovery of the sports activity level was significantly higher in the
bone-patellar tendon-bone group (p = 0.01), the present study showed an
initial significant improvement over the preoperative condition, but the mean
preinjury activity level was not achieved in either group at two years. While
most of our patients were recreational athletes, 65% of the patients in the
bone-patellar tendon-bone group and 79% of the patients in the hamstring group
who were involved in vigorous (IKDC level-I or II) sports activities before
the injury were able to return to the same level. In only 10% of the patients
in both groups was the inability to return to the preinjury level of sports
related to the knee.
The groups demonstrated similarly low rates of patellar symptoms, with
asymptomatic patellofemoral crepitation found in 22% of the patients in the
bone-patellar tendon-bone group and in 23% in the hamstring group.
Nevertheless, kneeling discomfort was more frequent in the bone-patellar
tendon-bone group (62%) compared with the hamstring group (15%) (p < 0.01).
No correlation was found between kneeling discomfort and the presence or the
extent of skin hypoesthesias, even though the bone-patellar tendon-bone group
had a larger mean area of decreased sensitivity (p < 0.001).
While some believe in the biological advantages of a multistranded
hamstring graft compared with a bone-patellar tendon-bone
graft51,52,
it is well accepted that healing of the tendon to bone is more difficult to
achieve and requires more time (usually eight to twelve weeks) than does
healing of bone to bone (usually four to six
weeks)53-55.
The attachment zone is said to be also more physiological with the
bone-patellar tendon-bone graft with a regular chondral transition between
tendon and bone56,
whereas in the hamstrings a fibrous insertion is usually
obtained55,57.
The factors that may determine the strength and stiffness ofthe
tendon-fixation device-bone complex after implantation are the tendon
graft-tunnel interface and the fixation device itself. A recent study in dogs
has demonstrated that pullout strength was enhanced by increasing the length
and the press-fit of the tendon within the tunnel. With doubling the length of
the tunnel, there was a 60% gain in terms of load to
failure58. Another
study in sheep59
has shown that the strength and stiffness of the tendon graftfixation complex
in the tibia was either maintained or improved with a low-profile distal
fixation device such as the WasherLoc screw. Therefore, for our patients in
the hamstring group, we increased the tendon-bone tunnel interface, adding
bone graft inside the tunnel, and we fixed the graft with the WasherLoc
screw.
We found the One Step tibial guide to be simple to use and reliable in
achieving a satisfactory position of the tibial tunnel. This guide allows the
surgeon to customize the tibial tunnel on the basis of the anatomy of the
individual patient. No patient had severe impingement. A small anterior
notchplasty is sometimes required. We avoided extending the notchplasty to the
posterior part of the notch because doing so might change the insertion point
of the graft and produce an abnormal graft tension pattern with knee
flexion60. The
amount of notchplasty is not accurately shown on postoperative
radiographs61—i.e.,
grafts that appeared to have impingement actually were free of
impingement—which may explain why the moderate impingements found were
of no consequence with regard to stability or motion. One advantage of the
bone-patellar tendon-bone graft over the hamstring graft is the ability to
rotate it to avoid impingement against the roof. In fact, the surgeon can
adjust the rotation of the bone-patellar tendon-bone graft on the femoral side
to position it more posteriorly and on the tibial side to reduce the
impingement. This cannot be done with the hamstring graft.
The position of the femoral tunnel is another important influence upon
graft tension during knee motion. In the sagittal plane, a posterior position
along the notch is preferable, but recently several
studies30,40
have emphasized the importance of the position of the femoral tunnel in the
coronal plane as well. It has been shown that a position at about twelve
o'clock, resulting from drilling the femoral tunnel through a vertical tibial
tunnel, can cause impingement of the graft and high graft tension in flexion.
To minimize these problems, the angulation of the tibial tunnel in the coronal
plane in this series averaged 66° in the bone-patellar tendon-bone group
and 69° in the hamstring group. The resultant femoral tunnel was in the
eleven o'clock position in the right knee and in the one o'clock position in
the left
knee62.
The fixation techniques for the hamstring graft that we adopted have been
tested
biomechanically20-23,27.
The Bone Mulch screw in the femur is a very strong and stiff fixation
device20. The four
strands of tendon remain parallel and can be tensioned
equally46 after
cycling. The femoral tunnel is grafted with autologous bone in order to
increase stiffness and avoid the pulley effect around the nose of the screw.
In the tibia, after grafting the tunnel, we used the WasherLoc system, which
also has shown good strength, stiffness, and slippage
characteristics21,60.
In the bone-patellar tendon-bone group, transcondylar screw fixation in the
femur was chosen to resemble the surgical technique used in the hamstring
group. The effectiveness of this method was recently reported in a two-year
anterior cruciate ligament reconstruction
trial26.
Interference screw and Tunneloc screw fixation were compared. Substantially
better knee stability and IKDC scores were found in those patients treated
with the Tunneloc screw. Furthermore, transcondylar fixation offers other
advantages, such as the absence of intraarticular hardware and greater
bone-to-bone contact surface, and it allows graft fixation in the case of
breakage of the posterior femoral wall.
The use of strong and stiff fixation allowed us also to achieve comparable
clinical results in men and women. Several
studies14,63,64
have described inferior results with the use of hamstring tendons in women,
but the fixation devices used in those studies had inferior biomechanical
properties. Moreover, in our series, the use of the One Step tibial guide, the
Bone Mulch screw, and the WasherLoc fixation system allowed us to accommodate
for the generalized laxity and the decreased bone mineral density often found
in women.
A certain amount of tunnel widening became apparent in a few months, most
often in the hamstring group and particularly in the femur. While the cause of
tunnel widening remains controversial and
multifactorial41,65,66,
it did not correlate with increased laxity in this study.
In conclusion, we believe that with modern surgical and fixation techniques
the same clinical results can be obtained with use of the two grafts. At the
present time, it is not possible to clearly show that one graft is superior to
the other. Specific indications for each of the two grafts have been presented
in the past, depending on the level of activity, body habitus, gender, and the
degree of joint laxity. From our results, the choice of the graft should not
be made on the basis on these criteria but on the patient's preferences and on
the surgical technique in which the surgeon is skilled. Surgeons who perform
many anterior cruciate ligament reconstructions need to master both
techniques. It is probable that the principles of surgical technique, graft
fixation, and postoperative rehabilitation are more important than the graft
choice in anterior cruciate ligament reconstruction.