Laboratory
studies1 have
suggested that the sagittal displacement that is permitted by a normal
knee2 or a total
knee replacement is influenced by the slope of the tibial
component3,4.
The less constrained the implant, the more pronounced the influence of slope
on anteroposterior displacement. If the prosthesis is nonconstrained, the knee
depends on the soft tissues and ligaments for anteroposterior stability;
consequently, a lesser degree of posterior slope and an intact anterior
cruciate ligament should be
beneficial3,4.
The posterior cruciate ligament usually is present and is without notable
degeneration in most osteoarthritic knees. In contrast, the anterior cruciate
ligament is frequently degenerated even if it is clinically functional at the
time of surgery. Some authors have suggested that an intact anterior cruciate
ligament is necessary for good short-term results after unicompartmental
arthroplasty5-9.
However, others have not found that a torn anterior cruciate ligament is a
contraindication to unicompartmental
arthroplasty10.
The purpose of the present study was to assess the effect of the posterior
slope of the tibial implant on the long-term outcome of unconstrained
unicompartmental arthroplasty in knees with intact and deficient anterior
cruciate ligaments.
We implanted 212 unicompartmental prostheses in 186 patients with
osteoarthritic knees between 1978 and 1988. One hundred and two patients (102
knees) died with the prosthesis intact before a minimum duration of follow-up
of twelve years and eleven patients (eleven knees) were lost to follow-up,
leaving seventy-three patients (ninety-nine knees) available for study. The
mean age of the patients at the time of surgery was seventy years (range,
forty-three to eighty-three years). Forty-nine right knees and fifty left
knees underwent unicompartmental arthroplasty. The medial compartment was
replaced in seventy-four knees. The implant that was used was the Lotus (Mark
1) unicompartmental prosthesis, developed by the Guepar Group (Howmedica,
Benoist Girard, Herouville Saint Clair, France), with a flat polyethylene
tibial component. At the time of the most recent follow-up evaluation,
twenty-two knees in fifteen patients had undergone revision
(Fig. 1). The reasons for the
revisions included loosening of the tibial implant (seventeen knees), wear in
the opposite compartment (three knees), infection (one knee), and patellar
impingement (one knee).
The ninety-nine knees in the present series (including the seventy-seven
knees that had not been revised and the twenty-two knees that had been
revised) were studied to evaluate the influence of the posterior slope of the
tibial implant on the long-term outcome of unicompartmental arthroplasty. The
seventy-seven knees (fifty-eight patients) that had not been revised were
evaluated clinically after a mean duration of follow-up of sixteen years
(range, twelve to twenty-four years). Anteroposterior stability was graded as
normal when there was =5 mm of anterior translation. We defined anterior
translation of >5 mm as a positive anterior drawer sign.
We defined the slope of the tibial implant as the angle in the sagittal
plane between the posterior inclination of the tibial implant and a line
perpendicular to the posterior tibial cortex
(Fig. 2). A positive value
indicated a posterior slope. Tibial translation on standing lateral
radiographs was measured at the time of the most recent follow-up with use of
the method of Dejour and
Bonnin2, which was
modified for the investigation of prosthetic knees. In a healthy
weight-bearing knee, tibial translation can be determined on a simple lateral
single-leg-stance
radiograph11
because in that anatomical reference position the posterior tangents to the
tibia and femur are
aligned12. In a
knee with a unicompartmental prosthesis, alignment of the femoral component
with the tibial component during weight-bearing may be altered by the position
of the components in the sagittal plane. Therefore, we used the displacement
of the nonimplanted femoral condyle with the posterior edge of the
nonresurfaced tibial plateau as the reference line
(Fig. 2).
Serial single-leg-stance weight-bearing lateral radiographs of the knee
were made during the follow-up period and at the time of the most recent
follow-up examination. Because the patients were elderly, the contralateral
limb was placed forward with only minimal weight-bearing to steady the
patient. As rotation may influence the measurement of displacement, the
radiographs were made with fluoroscopic guidance. Rotation was assessed
according to the position of the fibular head with respect to the tibia on the
radiograph and according to the position of the foot on the floor. When this
rotation was <20°, no compensation was used. For fourteen patients with
>20° of rotation, compensation was estimated with use of a
trigonometric formula. The postoperative alignment of the knee in the coronal
plane was evaluated as the hip-knee-ankle angle on standing radiographs of the
whole limb. Measurements were also made on anteroposterior and lateral
radiographs to determine the postoperative inclination of the femoral and
tibial components in the coronal and sagittal planes.
The outcome criteria at the time of the most recent follow-up included the
range of motion and the anteroposterior stability of the knee as assessed with
clinical examination, the anterior tibial translation as measured on standing
lateral radiographs, the number of knees with loosening of the tibial implant,
and the status of the anterior cruciate ligament in the knees that had been
revised. Multiple linear regression analysis was used to determine differences
in the posterior slope of the tibial implant between knees with and without
implant loosening at the time of the most recent follow-up while adjusting for
any confounding effects of age, weight, gender, thickness of the implant,
alignment of the knee, position of the implant, and status of the anterior
cruciate ligament. The Spearman test and multiple linear regression analysis
were used to determine possible correlations between specific variables. The
Mann-Whitney U test was used to identify the significance of the differences
between groups. Significance was defined as p < 0.05.
At the time of the operation, the anterior cruciate ligament was without
notable degeneration or other abnormality and was considered to be normal in
fifty knees (thirty-four of which had a medial arthroplasty and sixteen of
which had a lateral arthroplasty). The ligament was partially deteriorated (or
degenerated), had evidence of rupture of many fibers, and was considered to be
damaged but functional (as demonstrated by normal findings on anteroposterior
stability testing) in thirty-one knees (twenty-two of which had a medial
arthroplasty and nine of which had a lateral arthroplasty). The anterior
cruciate ligament was noted to be absent at the time of surgery in eighteen
knees (all of which had a medial arthroplasty). The posterior cruciate
ligament was present and appeared normal in all ninety-nine knees. There were,
therefore, three groups of knees that differed mainly with respect to the
status of the anterior cruciate ligament. During the period of implantation in
this series (1978 to 1988), the surgeons had a target posterior slope of
5° for the position of the tibial implant in the sagittal plane. The mean
posterior slope of the tibial implant of these ninety-nine knees was 5.1°
(range, —6° [anterior slope] to 18°).
Unrevised Knees
Among the seventy-seven knees that had not been revised at the time of the
most recent follow-up, thirty-seven had had a normal anterior cruciate
ligament at the time of implantation, twenty-nine had had a damaged anterior
cruciate ligament, and eleven had had no anterior cruciate ligament.
Anteroposterior stability was normal or nearly normal (=5 mm of anterior
displacement) in sixty-six knees. There was a positive anterior drawer sign
(>5 mm of anterior displacement) in the eleven knees in which the anterior
cruciate ligament had been absent at time of surgery. With the numbers
available, there was no relationship between the tibial slope of the implant
and the maximum flexion angle of the knee or the flexion deformity of the knee
at the most recent follow-up (p = 0.21 and p = 0.32, respectively; Spearman
test).
The mean anterior tibial translation on weight-bearing radiographs was 3.7
mm (range, —2 to 10 mm), and the mean posterior slope of the tibial
implant was 4.3° (range, —6° [anterior slope] to 10°). There
was a significant relationship between anterior tibial translation as seen
radiographically and posterior tibial slope (Rs = 0.402; p < 0.01). This
correlation between anterior tibial translation and posterior tibial slope was
found regardless of whether the anterior cruciate ligament was present (p =
0.005) or deficient (p = 0.001). The mean posterior slope of the tibial
implant was 0° (range, —6° to 4°) among the eleven knees in
which the anterior cruciate ligament had been absent at the time of
implantation, compared with 4.7° (range, —4° to 10°) among
the sixty-six knees in which the anterior cruciate ligament had been present.
There was a significant difference in tibial translation between the knees
with a normal or damaged anterior cruciate ligament and those without an
anterior cruciate ligament (p < 0.01). The mean tibial translation at the
time of follow-up was 6.9 mm (95% confidence interval, 5 to 9 mm) for the
eleven knees in which the anterior cruciate ligament had been absent at the
time of surgery, compared with 3.2 mm (95% confidence interval, 0 to 6 mm) for
the sixty-six knees in which the anterior cruciate ligament had been present.
No significant difference was observed, with the numbers available, between
knees with a normal anterior cruciate ligament and those with a damaged
anterior cruciate ligament.
Revised Knees
Posterior Slope and Revision for Loosening of the Tibial Implant
(Table I)
By the time of the most recent follow-up, seventeen knees had been revised
for loosening of the tibial implant. At the time of implantation, eight knees
had had a normal anterior cruciate ligament, two knees had had a damaged
anterior cruciate ligament, and seven knees had had no anterior cruciate
ligament. At the time of the revision for loosening, only five knees had an
anterior cruciate ligament; three of these ligaments had been normal at the
time of implantation, and two had been damaged. The mean posterior slope of
the tibial implant in these five knees was 0° (range, —5° to
3°) (Table I). Five other
knees that had had a normal anterior cruciate ligament at the time of
implantation had no anterior cruciate ligament at the time of revision. The
mean posterior slope of the tibial implant in these five knees was 14°
(range, 13° to 18°). The mean posterior slope of the seven tibial
implants that had been revised for loosening in knees in which the anterior
cruciate ligament had been absent at the time of implantation was 11°
(range, 9° to 12°) (Table
I).
When multiple linear regression analysis was used to remove the confounding
effects of age, weight, gender, alignment of the knee, and the thickness and
position of the implants, the mean posterior slope of the tibial implant was
significantly less in the group of knees without loosening of the tibial
plateau than in the group of knees with loosening (p = 0.034).
Status of the Anterior Cruciate Ligament at the Time of Revision
Five implants were retrieved for reasons other than loosening, including
infection, impingement of the patella, and involvement of the opposite
tibiofemoral compartment. The anterior cruciate ligament was present at the
time of revision in all five knees, and the mean posterior slope of the tibial
implant was 1° (range, —2° to 4°). Therefore, ten knees that
were revised (including five with loosening and five without) had an intact
anterior cruciate ligament at the time of revision. Two of these ten knees had
had a damaged anterior cruciate ligament at the time of implantation, but the
ligament still appeared to be functional at the time of revision. The other
eight knees had had a normal anterior cruciate ligament at the time of
implantation. In three of these eight knees, the synovial covering of the
ligament was still present at the time of the most recent follow-up. In the
other five knees, the synovial covering of the ligament had been lost and the
ligament had been split longitudinally into separate bundles of fibers; these
fibers were discolored, suggesting that the ligament had deteriorated. The
posterior cruciate ligament was present and appeared normal in all knees at
the time of revision.
The anterior cruciate ligament had been present at the time of
unicompartmental arthroplasty in eighty-one of the ninety-nine knees in the
present series. Of the eighteen knees in which the anterior cruciate ligament
had been absent at the time of implantation, eleven still had the implant in
situ at the time of the most recent follow-up at a mean of seventeen years
(range, fifteen to twenty-four years) postoperatively. In these eleven knees,
the posterior tibial slope was <5° (mean, 0°; range, —6°
to 4°) and limited anterior tibial translation (<10 mm) was seen on
lateral radiographs (Table I).
In the seven knees without an anterior cruciate ligament at the time of
unicompartmental arthroplasty that were subsequently revised, the posterior
tibial slope was >8° (mean, 11°; range, 9° to 12°) and
anterior tibial translation of >10 mm was observed on lateral radiographs
that were made at the time of revision for tibial loosening
(Fig. 3)
(Table I).
We believe that tibial slope is an important factor that controls tibial
translation during weight-bearing in association with this unconstrained
implant design. Posterior tibial slope is important for knee stability in the
sagittal
plane2,13.
Dejour and Bonnin found that, regardless of the condition of the anterior
cruciate ligament, every 10° increase in the posterior inclination of the
tibial plateau was associated with a 6-mm increase in anterior tibial
translation in monopodal
stance2. This
phenomenon was also observed in our series. With regard to tibial translation
during weight-bearing, we found that the unconstrained unicompartmental
prosthesis behaved similarly to a healthy
knee2,12.
The magnitude of tibial slope affects the stabilizing effect of the anterior
cruciate ligament. An increase in the posterior tibial slope was associated
with an increase in tibial translation, even when the anterior cruciate
ligament was present. However, the translation was much more pronounced in
cases in which the anterior cruciate ligament was deficient. It should be
noted that the polyethylene tibial implant of this unicompartmental prosthesis
is flat (maximally nonconstrained) and that the results could be different
with a more congruent polyethylene tibial implant.
An increased posterior tibial slope has often been recommended to promote
increased flexion and femoral
rollback1 and to
improve the stress distribution at the bone-tibial component interface. In a
laboratory study, Garg and
Walker1 found a
significant improvement in mobility when the posterior slope reached 10°.
In contrast, our clinical findings suggest that an increased posterior slope
does not improve mobility and may cause major translational movements even if
the anterior cruciate ligament has been retained. We view a posterior slope of
between 3° and 7°, as recommended by Whiteside and
Amador14, as a
compromise between the risk of excessive translation on the one hand and the
risk of excessive stress and even cruciate ligament avulsion on the other.
Many
authors5-9
have suggested that an intact anterior cruciate ligament is necessary in order
to achieve a good result following unicompartmental arthroplasty, but the
natural history of
gonarthrosis15 is
associated with degeneration of the anterior cruciate ligament. Can the
anterior cruciate ligament be preserved for ten or twenty years in patients
managed with unicompartmental
arthroplasty16? In
the group of eighty-one knees in which the anterior cruciate ligament had been
present at the time of implantation, five ruptures occurred by the time of the
most recent follow-up. In these five knees, disruption of the anterior
cruciate ligament was not associated with radiographic evidence of progressive
deterioration of the articular cartilage in the patellofemoral joint or in the
nonresurfaced femorotibial compartment. There were no large femoral
intercondylar or tibial osteophytes to suggest that a progressive degenerative
process had altered the ligament. Rather, the implant in all five knees had a
posterior slope of >10°. The anterior tibial translation increased from
the first postoperative year to the time of revision in these five knees,
suggesting that progressive disruption of the anterior cruciate ligament
occurred over time in relation to the increased posterior slope of the tibial
component.
We were able to inspect the anterior cruciate ligament in ten knees during
revision surgery at a mean of fourteen years (range, six to twenty years)
after implantation of the prosthesis. In some of these knees, we observed loss
of the synovial covering of the anterior cruciate ligament at an early stage
and longitudinal splitting of the ligament into separate fiber bundles. These
bundles were discolored and unnaturally friable, suggesting that the ligament,
in some circumstances, may deteriorate for other than mechanical reasons.
However, in the absence of an increased posterior slope, none of the anterior
cruciate ligaments with signs of deterioration ruptured and all still appeared
to be functional at the time of revision. We carried out wide excision of
osteophytes to relieve so-called notch stenosis at the time of the
unicompartmental
arthroplasty17. No
notch stenosis due to osteophytes was found in those knees at the time of
revision.
In the present series, a functional anterior cruciate ligament was present
in eighty-one knees at the time of surgery and in seventy-six knees at the
time of the latest follow-up. A comparison of the results for the knees that
had a normal-appearing anterior cruciate ligament and those that had partial
degeneration of the anterior cruciate ligament at the time of implantation
showed no significant difference, with the numbers available, in terms of
either anteroposterior stability on clinical examination or anterior tibial
translation as measured on radiographs. Even knees that showed partial
degeneration of the anterior cruciate ligament at the time of surgery had good
anterior drawer stability at the time of the most recent follow-up. Therefore,
while there is little direct evidence with regard to the fate of the anterior
cruciate ligament after unicompartmental replacement, we suggest that avoiding
too great a posterior slope protects the ligament from degeneration and
rupture, which might otherwise eventually occur in the natural course of the
disease.