A supracondylar varus osteotomy of the femur can be performed
to correct a valgus deformity that is secondary to osteoarthritis
of the lateral compartment of the knee. The goal of treatment is
to obtain axial correction without extensive dissection. The postoperative
regimen should impose the least possible restriction on the lifestyle
and daily activities of the patient.
Osteosynthesis provides a means for obtaining maximum stability
of the osteotomized femur and achieving the goals of the osteotomy.
Müller et al.12, in 1970, introduced
the revolutionary concept of obtaining stability at the osteotomy
site with use of 90-degree double-angle and 95-degree condylar plates.
This method was widely adopted on the basis of the concept that,
with use of these stiff implants, early postoperative walking with
partial weight-bearing could be instituted. However, despite rigid
fixation of the osteotomy site, the cortical bone of the proximal
segment often subsided into the cancellous bone of the distal segment when
the patient was bearing weight, resulting in unwanted axial deviation16. This occurred frequently in osteoporotic
bone or after repeated intraoperative impaction of the seating chisel,
to obtain the desired correction, resulting in loose seating of
the blade.
These concerns led some authors to avoid vigorous postoperative
rehabilitation. Finkelstein et al.4 immobilized
the extremity in a cylinder cast for two weeks and prohibited weight-bearing
for six weeks. McDermott et al.9 permitted
walking with graduated weight-bearing only when radiographs showed
signs of consolidation. Healy et al.5 stabilized
the leg with a hinged fracture-brace and allowed only touch-down
weight-bearing for six weeks.
Since 1984, an incomplete oblique supracondylar osteotomy with
use of a malleable semitubular plate, bent to form an angled plate,
has been performed at our hospital to achieve stabilization. Postoperatively,
the patient is immediately allowed to walk with crutches while bearing
partial weight on the affected extremity. We reviewed the results
of this treatment to determine whether the osteosynthesis could
withstand the forces generated during functional postoperative rehabilitation and
whether the planned angle of correction could be achieved and maintained
until there was bone-healing.
Between 1984 and 1996, we performed an oblique supracondylar
osteotomy of the distal part of the femur in twenty-one knees in
nineteen patients (nine women and ten men). There were ten right knees
and eleven left knees; two patients had bilateral involvement, and
the second operation was performed more than six months after the
first procedure. At the time of the index operation, the mean age
was fifty-seven years (range, thirty-nine to seventy-one years).
None of the patients were lost to follow-up.
Preoperatively, all patients complained of pain that was predominantly
localized to the lateral side of the knee and was typically present
when climbing stairs or walking on uneven ground. Two patients had
to change their occupation, because of moderate or severe pain,
from a physically demanding job involving manual labor to a sedentary job.
All patients had to reduce their physical activity, and most patients
used nonsteroidal anti-inflammatory medications to reduce the pain.
Radiographs showed mild or moderate osteoarthritis of the lateral
compartment of the knee, with narrowing of the lateral joint space,
subchondral sclerosis, and few if any osteophytes. The mean tibiofemoral
angle was 12 degrees (range, 10 to 16 degrees). The mean knee flexion
was 148 degrees (range, 130 to 160 degrees); four patients had a slight
extension deficit, and one patient had a 15-degree extension deficit
(Table I).
The indications for the procedure included pain secondary to
mild or moderate osteoarthritis and a valgus deformity with a tibiofemoral
angle of up to 20 degrees. Patients also had to be able to comply with
a program of partial weight-bearing with use of crutches. The degree
of pain was determined by questioning the patients rather than by
using a visual analog scale. Contraindications to the procedure
included narrowing of the medial compartment of the knee, tibiofemoral
subluxation, patellofemoral arthritis, valgus deformity due to obliquity
of the tibial plateau, an extension deficit of more than 15 degrees,
instability due to laxity of the medial collateral ligament, and
severe osteoporosis.
Postoperatively, the functional outcome was evaluated with use
of The Hospital for Special Surgery knee-rating scale6; this was the scale that was commonly
used at the time that the senior author (F. H.) first performed the
index procedure. The scale has a maximum of 100 points, with 30
points assigned to pain; 22 points, to function; 18 points, to range
of motion; and 10 points each, to muscle strength, an extension
deficit, and instability. Points are deducted for use of walking
aids, an extension lag, and varus deformity. The overall score is
categorized as a failure (less than 60 points), fair (60 to 69 points), good
(70 to 84 points), or excellent (85 points or more). Subjective
satisfaction with the outcome of the procedure was determined by
asking the patients whether, given the same preoperative conditions,
they would have the same operation again.
The effects of mechanical loading of the osteosynthesis were
evaluated in relation to compliance with the postoperative regimen
(the time at which the patient stopped using crutches and use of
external immobilization). Compliance with the postoperative regimen
was confirmed by the senior author, who observed all of the patients
during physiotherapy, which was provided at the hospital. Observation
of the patient's ability to bear partial weight, and the patient's
description of his or her activities, were useful for estimating
the effects of mechanical loading and also for monitoring and guiding
the patient's progress. Patients were questioned about pain during
weight-bearing, since pain can be considered a sign of instability.
The radiographs made immediately after the operation were compared
with those made at eight weeks and one year postoperatively, in
order to determine whether subsidence of the proximal cortex into
the distal cancellous bone or signs of implant loosening were present.
We evaluated whether there was any difference between the planned
tibiofemoral angle of correction to 1 to 3 degrees of varus (depending
on the height of the patient) and the final angle of correction
after bone-healing. The planned correction was predetermined on
a tracing of the preoperative radiograph. The final correction that
was achieved after bone-healing was determined from the radiograph
used for preoperative planning and the radiograph made one year
after the operation, when the patients could walk without crutches
or pain while bearing full weight on the affected leg. Use of these
criteria ensured that the femur had healed and was stable at the
time that the one-year follow-up radiograph was made. The anteroposterior
radiograph that was used to determine the angle of correction was made
with an eighteen-by-forty-three-centimeter x-ray cassette with the
patient standing. The patella was carefully centered on the distal
aspect of the femur between the two epicondyles, especially in patients
who had an extension deficit. All of the preoperative and postoperative
radiographic measurements were made by one individual who was unaware
of the initial planned angle of correction. All of these measurements
were estimated independently by another orthopaedic surgeon who
did not participate in the study.
Operative Technique
The operation was performed without use of a tourniquet, through
a medial skin incision that extended ten centimeters proximal to
the medial epicondyle of the femur. The vastus medialis muscle was
freed from the intermuscular septum, and periosteal branches from
the superior medial genicular artery were ligated. The periosteum
was incised longitudinally and retracted with Hohmann retractors
to expose the cortex at the planned level of the osteotomy.
The level and direction of the osteotomy were determined with
use of a small radiopaque triangular template that was available
in several different angles. Image intensification was used to confirm
the placement of the template, with care being taken to obtain a
true anteroposterior image by centering the patella between the
two epicondyles. The template corresponding to the size of the bone
wedge to be removed was placed frontally over the femur with its
tip pointing toward the lateral epicondyle, with care taken to ensure
that the tip remained proximal to the joint surface. The template
was rotated around its tip until the sides crossed the medial cortex
at a point equidistant from the tip (Fig. 1-A). The point of entry for the plate
was marked on the medial epicondyle with a Kirschner wire. The flexible
target triangle was held against the curvature of the bone while
the lines of the osteotomy were marked along the sides. The template
was removed, and the osteotomy of the femur was performed along
the markings, with care taken to leave the lateral cortex intact.
After the bone wedge had been removed, a slight varus force was applied
across the defect for a few minutes in order to slowly bend the
lateral cortex without breaking it. On completion of the osteotomy,
the medial cortices of the two segments of bone should be exactly
apposed to each other.
Osteosynthesis was achieved with use of a six-hole semitubular
steel AO plate (Stratec Medical, Oberdorf, Switzerland). One-half
of the plate was beaten flat with a hammer; the implant then was bent
through the third screw-hole to form an angled plate, with the flat
portion forming the blade. The newly fashioned blade was driven
into the condyle two centimeters distal and approximately parallel
to the osteotomy line (Fig. 1-B). A cortical lag-screw was placed
through the hole at the bend and was angled in an ascending and slightly
posterior direction. The screw could pass through the plate into
the thick lateral cortex in an oblique direction and obtained excellent
purchase in the bone, thereby permitting compression of the osteotomy
site (Fig. 1-C).
The plate was secured proximally with three cortical screws, which
were inserted eccentrically; this placed the plate under tension
and allowed it to conform to the shape of the bone, thereby further
compressing the osteotomy site (Fig. 1-D). In five patients, the bone appeared
less dense than expected and a second lag-screw with a washer was
placed anterior to the plate to obtain more compression laterally
and to prevent rotation of the device (Fig. 1-E,Fig. 2-A and Fig. 2-B). We found
use of the second screw to be advantageous in patients who had osteopenic
bone.
Postoperative Regimen
Active range-of-motion exercises were started on the first postoperative
day. The patients were allowed to get out of bed and were instructed
to use crutches while bearing about twenty kilograms of weight on
the involved extremity during walking. This degree of weight-bearing
was maintained for eight weeks, during which time swimming or early return
to a standing occupation were not permitted. Subsequently, if there
was no evidence of subsidence at the site of the osteotomy or of
loosening of the implant and there was no pain, full weight-bearing
was permitted, even if there was little evidence of callus formation
on the radiographs. There was usually no evidence of callus formation at
eight weeks postoperatively.
Function
The mean overall preoperative knee score, according to the scale
of The Hospital for Special Surgery6,
was 65 points (range, 56 to 70 points) for the twenty-one knees.
This score increased to a mean of 84 points (range, 61 to 100 points)
at a mean of five years (range, two to twelve years) postoperatively
(Table II).
Most of the increases were due to a reduction in pain compared with
the preoperative status. Eleven knees were rated as having an excellent
result; eight, a good result; and two, a fair result. One patient
who had a fair rating had more pain during walking nine years after
the operation (and after a long symptom-free period) than she had
had preoperatively. The other patient who had a fair rating had
a rapid decrease in function secondary to dementia six years after
the operation. The osteosynthesized femur healed in twenty knees.
One patient had failure of healing. His score was 80 points three
years after a second osteosynthesis procedure.
One of the two patients who had had to change from a job involving
manual labor to a sedentary occupation preoperatively returned to
heavy farm labor thirteen months after the operation. The other
patient was free of symptoms after the operation but preferred to
continue driving a bus, for personal reasons.
Twelve patients had a high level of subjective satisfaction with
the outcome. One additional patient was very satisfied with the
result in one knee at twelve years postoperatively but was only
moderately satisfied with the result in the other knee at eleven
years. Four patients were moderately satisfied. One patient was
dissatisfied eleven years after the procedure because of increasing
pain during the two years prior to the latest evaluation. The remaining
patient could not describe her degree of satisfaction because of
dementia; however, her daughter stated that her mother had performed household
tasks for many years after the operation and had never mentioned
any problems related to the involved extremity. Eighteen patients
stated that they would have the operation again given the same set
of circumstances. No patient required conversion to a total knee
replacement at the time of the latest follow-up evaluation.
Compliance with Postoperative Regimen
Fourteen patients stopped using crutches ten weeks or less after
the procedure, after having been fully weight-bearing for a short
period. An additional patient, who had been managed with a bilateral
operation, stopped using crutches ten weeks after one operation
and 1.5 years after the other. Two patients used crutches for twelve weeks,
and the duration of use was not known for two patients.
Seventeen patients were treated without any form of external
immobilization. The two remaining patients were managed with immobilization
of the extremity in a cylinder cast for four and eight weeks as
a precaution; one of these patients had pain when bearing weight
on the affected extremity, and the other patient was unable to comply with
partial weight-bearing. A cast was preferred to a more functional
brace because the cost of the latter would have been borne by the
patient, whereas that of the former would not.
Pain As a Measure of Healing
Fifteen of the knees were not painful during weight-bearing.
However, most patients reported a feeling of weakness in the involved
extremity that lasted for about six months. Three knees were painful
during weight-bearing, and one was immobilized for this reason.
It was not clear whether the three remaining knees were painful
during weight-bearing.
Radiographic Findings
Postoperatively, there was no subsidence in eighteen knees. One
knee had slight subsidence on the medial side, with a negligible
change in the angle between the tangent to the condyles and the
longitudinal axis of the femur. One knee with a failed osteosynthesis
had subsidence on the lateral side. In another knee, it was not
possible to assess subsidence because the first postoperative radiograph could
not be retrieved.
Nineteen knees had no signs of implant loosening, and two had
loosening. In one of the latter two knees, the lag-screw crossing
the osteotomy site broke without resulting in any loss of correction.
For eighteen knees, the mean difference between the planned correction
of the longitudinal axis of the extremity and the final correction
achieved after bone-healing was 1.7 degrees (range, 0 to 4 degrees).
The interobserver difference was 0.04 degree for the mean difference
and 0 degrees for the maximum difference. For two knees, the preoperative
radiographs could not be retrieved; however, the follow-up radiographs
showed no subsidence or loosening. The follow-up radiograph was
not available for the remaining knee (Table II).
Complications
The patient in whom the osteosynthesis failed was a healthy fifty-seven-year-old
farmer. She complained of increasing postoperative pain during weight-bearing,
mainly on the lateral side of the knee. The pain was initially attributed
to an excessively long screw, which was removed eight weeks after
the operation. At that time, there were no radiographic signs of
instability; however, the blade of the plate was less than two centimeters distal
to the osteotomy site. The leg had not been immobilized, as the
patient refused to wear a cast. Removal of the screw did not reduce
pain during weight-bearing as much as had been expected. Four months
after the operation, a progressive valgus deformity developed, with
subsidence of the proximal cortex into the distal cancellous bone
on the lateral side and loosening of the screws. A second procedure
was performed with use of a double-angled plate, without any intraoperative problems.
The osteotomy site healed uneventfully, and the patient did well
postoperatively.
Minor complications included two subcutaneous hematomas, noted
on the second and third postoperative days, which resolved without
treatment; one popliteal vein thrombosis, noted on the fourth postoperative
day, which responded to a ten-week course of Coumadin (warfarin);
and one superficial wound infection, noted on the fourth postoperative
day, which resolved after use of antibiotics. None of these complications
led to a longer period of rehabilitation.
It is believed that the exogenous stability achieved with stiff
implants can be enhanced by use of inherent endogenous stabilizing
mechanisms of the body. Wagner16 and,
later, Miniaci et al.11 tried
to improve the congruence of the superimposed cortices by making
the bone cut in a descending direction toward the condyle. Others attempted
to keep the opposite cortex and the periosteum intact, by only bending
it, so that it would act as a stabilizing hinge1,3,7,15,16.
However, oblique and incomplete osteotomies have been used mainly
for correction of varus deformity of the proximal aspect of the
tibia13,17. The use of implants
that are shorter and have a lower profile than the double-angle
and condylar plates made it possible to plan a less extensive operation.
Furthermore, smaller implants may not require removal. Weber and
Wörsdörfer18 and Wagner16 used a malleable semitubular plate,
which can easily be contoured to the bone, with one end trimmed
to form a serrated edge that can be driven into the femoral condyle.
However, low-profile implants have never been widely used because
of concern that they would provide decreased stability compared
with that afforded by stiff implants.
In seventeen of the twenty-one knees, the osteosynthesis accomplished
with our method withstood the mechanical loads generated during functional
postoperative rehabilitation. In three knees, the osteosynthesis
was stable but required a prolonged period of partial weight-bearing
or immobilization, or both. Osteosynthesis failed only in one knee
(5 percent). In comparable studies in which stiff, angled implants
were used, mechanical complications related to failure of fixation were
reported in 4 to 16 percent of patients1,2,4,5,9-11,14.
We believe that the advantage of our technique lies in the stability
offered by the fixation, which is based on two factors. First, the
descending direction of the osteotomy leads to complete circumferential
abutment of the superimposed cortical segments because both surfaces
have almost the same diameter. This degree of circumferential congruence
cannot be achieved with a horizontal osteotomy (Fig. 3). In contrast
to contact between the proximal cortex and the distal cancellous
bone, circumferential contact between the two cortical surfaces
at the osteotomy site renders subsidence with subsequent deviation
of the axis much less likely. Second, the stability of an interrupted
tubular structure can be reinstated by frictional forces when the
two parts of the tube are sufficiently compressed onto each other.
The entire contact area of the superimposed cortices is compressed
by the lag-screws crossing the osteotomy site as well as by the
screws placed eccentrically in the plate.
Another advantage of our technique is its potential to achieve
the desired correction of the longitudinal axis of the extremity.
The average deviation of 1.7 degrees between the planned and final
angles of correction after bone-healing suggests that this technique
allows the desired degree of correction to be achieved. Our available
radiograph size (eighteen by forty-three centimeters) renders measurement
of the longitudinal axis of the extremity more difficult than it
is on a radiograph made with a thirty-six-inch (ninety-one-centimeter)
cassette. However, by paying meticulous attention to details and
ensuring that all radiographs and measurements were made with use
of a standardized method, we were able to avoid errors due to the closely
placed reference points.
There are two reasons why such accurate correction can be achieved
with our technique. First, the intraoperative correction of the
axis of the extremity is determined according to the size of the wedge
of bone that is removed. With a malleable implant, the direction
of the plate in the condyle has no impact on the extent of the correction
because tightening of the screws will contour the plate to the bone
rather than pulling the bone to the plate. Only a few approximated
reference points are required to achieve accurate correction. The degree
of correction is not affected by the slope of the wedge of bone.
In contrast, with a fixed-angle device, correction of the axis of
the extremity is determined mainly by the direction in which the seating
chisel is driven; the size of the wedge of bone that is removed
plays a secondary role. With a fixed-angle device, subsidence of
the cortices into one another is even exploited to make adjustments
of the correction intraoperatively9.
The use of stiff implants makes it necessary to determine several
reference points, each introducing a potential source of error.
It seems far simpler to accurately remove the planned wedge of bone
than to maintain correct angulation in three planes while driving
a seating chisel into the condyle of the femur. The second reason
for the accuracy of the correction is that cortical tubes that abut
each other in a congruent manner are less likely to subside into
one another than are less precisely superimposed cortices. Therefore,
deviation of the axis under load is less likely after an oblique
osteotomy than it is after a horizontal osteotomy.
A difference of less than 2 degrees between the planned and final
angles of correction of the axis of the extremity has been reported
with use of stiff implants8,9.
However, the maximum difference between the planned correction and
the final result in comparable studies in which fixed-angle devices
were used was more than the 4 degrees observed in our patients.
McDermott et al.9 performed a
corrective osteotomy in twenty-three patients with use of guide-wires
to seat the chisel parallel to the joint space for correct orientation
of the 90-degree double-angle plate. Those authors reported 80 percent
concordance between the planned and final angles of correction.
However, the maximum difference between the planned and the final
correction was twice the difference seen in our patients. Learmonth8 used a special jig applied to the
lower extremity to identify the longitudinal axis of the knee and
reported a mean and maximum difference, between the planned correction
and the final result, of 0.3 and 2 degrees, respectively, in twelve
patients.
Maintenance of an intact lateral hinge is crucial for the success
of our technique. It is therefore important to close the bone wedge
slowly to allow the lateral cortex to bend. Although it was not
always possible to avoid a fracture of the lateral cortex, the experience
of the senior author has shown that the intact periosteum and joint
ligaments are sufficient to maintain stability in conjunction with the
osteosynthesis. Furthermore, the intact lateral structures prevent
rotational and translational malalignment until full stability has
been restored by the osteosynthesis.
We believe that the operative technique that we have described
provides an alternative to the conventional method of using a stiff
implant for the correction of mild-to-moderate valgus deformity of
the knee.
Note: The authors wish to thank Alberto G. Schneeberger, M.D.,
for repeating the measurements on the radiographs.