Curettage and cementation often are used in the treatment of benign
aggressive bone lesions such as giant-cell tumors. A large cortical window is
created over the tumor, which is removed with curettage, and the cavity is
enlarged with a high-speed burr. An adjuvant, such as liquid nitrogen, phenol,
or argon beam coagulation, sometimes is used to extend the margin of tumor
kill1,2.
Finally, the cavity is packed with acrylic bone cement.
The use of polymethylmethacrylate to fill cavitary defects was first
reported in 19693.
In 1976, Persson and Wouters proposed the use of acrylic cement to extend the
effective margins of a curettage by thermal
necrosis4. Of the
many bone fillers available, cement has some unique advantages. Filling the
cavity with cement helps to maintain the osseous anatomy, provides immediate
stability, and preserves the joint surface, allowing earlier physical
rehabilitation and return to activity. The plasticity of cement during
polymerization allows complete filling of a tumor cavity with irregular
confines, and the polymer can be firmly molded around metallic implants and
fixation devices. Tumor recurrence at the bone-cement interface is easily seen
on radiographs, and other treatment alternatives are not precluded by the use
of acrylic cement5.
Subchondral stiffness has been shown to be nearly 98% of that of the
contralateral limb when limb defects are reconstructed with
cement6.
A serious problem following extended curettage and cementation in the
treatment of giant-cell tumors has been the frequency of postoperative
fracture of the host bone. Marcove et
al.7 reported on
fifty-two patients with benign giant-cell tumors treated with extended
curettage, cryosurgery with liquid nitrogen, and cementation. Although the
rate of tumor recurrence was low, approximately 25% of the patients had
postoperative fractures. In an attempt to reduce the frequency of
postoperative fractures, Bini et
al.8 used
intramedullary Steinmann pins to augment the cement. No postoperative
fractures occurred in their thirty-eight patients.
In a biomechanical study, Randall et
al.9 used three
intramedullary Steinmann pins to augment the cement in noncontained defects in
the proximal part of the tibia. The pins were positioned so that the proximal
ends supported the tibial joint surface, while the distal ends achieved an
interference fit in the medullary canal. The use of intramedullary pins in the
cement demonstrated a mechanical advantage compared with the use of cement
alone. Three similar biomechanical
studies10-12,
however, concluded that the addition of intramedullary Steinmann pins to the
cement did not significantly improve the strength of the reconstruction for
contained defects.
The purpose of this biomechanical study was to examine an alternate method
of augmenting the polymethylmethacrylate used in reconstructing osseous
defects created by curettage of benign bone tumors in the distal aspect of the
femur. The cement was augmented with crossed screws that purchased healthy
bone on the opposite side of the defect, and this construct was compared with
constructs that used cement alone and cement with intramedullary pins.
Twenty pairs of matched femora from human cadavera were divided into
four groups of five pairs each. Group I contained five pairs of formalin-fixed
femora (University of Tennessee, Memphis, Tennessee). Groups II, III, and IV
each contained five pairs of fresh-frozen femora (International Institute for
the Advancement of Medicine, Phoenix, Arizona). The average age (and standard
deviation) of the donors at the time of death was 71.4 ± 5.37 years for
the formalin-fixed femora and 76.3 ± 5.92 years for the fresh-frozen
femora.
The femora in group I were harvested from formalin-fixed cadavera and were
stripped of all soft tissues. The femora from the remaining groups were
obtained as fresh-frozen specimens and had little soft tissue. Radiographs
were made of all femora to ensure osseous normality.
Specimen Preparation
After the fresh-frozen specimens were thawed at room temperature, a
high-speed burr (Stryker, Kalamazoo, Michigan) was used to create noncontained
defects in the medial femoral condyles of all specimens except for five of the
ten femora from group IV. The defect was designed to represent the cavity that
remains after curettage of a giant-cell tumor of the medial femoral condyle.
Cortical and cancellous bone was removed from the subchondral plate to the
metaphyseal-diaphyseal junction. The defect was extended laterally to the
midpoint of the trochlear groove. An effort was made to ensure that the
thickness of the subchondral plate was uniform and comparable between each
matched pair. The thickness of the subchondral plate was assessed by feel
alone. To minimize the effect of human inaccuracy and bias, the method of
reconstruction was determined by a random-number generator after both defects
had been created within each matched pair. All thirty-five defects were
created by the same individual and were kept moist during preparation and
testing.
Group-I (formalin-fixed) femora and group-II (fresh-frozen) femora were
used to compare the strength of reconstruction with use of cement alone and
reconstruction with use of cement augmented with crossed screws. In each pair,
one femoral defect was randomly assigned to receive cement and crossed screws
and the contralateral femur was assigned to receive cement alone. Screw
augmentation consisted of four divergent screws placed to gain purchase in the
intact remaining bone (Fig. 1).
A 3.2-mm bit was used to drill four holes that extended to the opposite side
of the femur. For two of the holes, the drill was directed from the distal
aspect of the defect into the cortical bone proximally. For the other two
holes, the drill was directed at an inferior angle through the cancellous bone
of the lateral condyle. Two 4.5-mm cortical screws were placed in the
superiorly directed holes, and two 6.5-mm cancellous screws were placed in the
inferiorly directed holes. Screw length was determined so that the screw heads
approximately marked the imaginary border of the normal medial condyle.
Sterile prepackaged polymethylmethacrylate (VersaBond; Smith and Nephew
Orthopaedics, Memphis, Tennessee) was prepared and hand-mixed according to the
manufacturer's instructions. Once the cement was nonadherent to gloves, it was
press-fit around the screws and into the defect. The cement was molded to
approximate the physiologic contour of the removed medial condyle. After the
curing process, the screw heads protruded slightly from the cement. The
contralateral femora in groups I and II were reconstructed with
polymethylmethacrylate alone.
Group-III fresh-frozen femora were used to compare crossed-screw
augmentation with intramedullary Steinmannpin augmentation. Again, the right
or left femur of each pair was randomly chosen to be reconstructed with cement
augmented with screws. The contralateral side was reconstructed with use of
cement augmented with 3/16-in (4.8-mm) threaded intramedullary Steinmann pins
(Smith and Nephew Orthopaedics). The pins were fanned out to provide good
support to the entire medial joint surface. As described by Randall et
al.9, the proximal
ends of the pins extended into the diaphysis and were manually placed to
achieve an interference fit in the medullary canal
(Fig. 2). The number of pins
chosen (four or five) depended on the size of the medullary canal. Cement was
then press-fit around the pins, molded to approximate the contour of the
medial femoral condyle, and allowed to cure.
Group-IV fresh-frozen femora were used to compare defects reconstructed
with cement augmented with crossed screws and femora without any modification.
The femora in this group underwent the same modifications and were
reconstructed in the same manner as previously described.
Mechanical Testing
All femora were transected 15 cm proximal to the joint line with use of a
handheld saw. Each femur was placed into a square reusable 316-L stainless
steel fixture and was rigidly fixed in cold-cure acrylic cement (Memphis
Dental Manufacturing, Memphis, Tennessee) at an angle that placed the
transcondylar axis horizontally when mounted in the testing apparatus.
Each femur and associated fixture were bolted to the load frame of an
Axial/Torsion Test System (Interlaken, Eden Prairie, Minnesota) with an 8900-N
load-cell. A loading nose with a flat surface area of 8 cm2 was
secured to the actuator. The loading nose was centered over the medial condyle
for each specimen (Fig. 3). The
femora were then tested in the manner described by Randall et
al.9. Beginning at
zero load and displacement, increasing force was applied at a rate of 10 N/sec
until a load of 475 N was achieved. The femora in groups II, III, and IV were
then cycled in a sinusoidal fashion between 50 N and 900 N for 2000 cycles at
1 Hz. Because formalin preservation decreases the energy absorption capacity
of bone, it was believed that group-I femora would not survive cycling. The
group-I femora, therefore, underwent only load-to-failure testing. The
actuator position at the initiation of load-control cycling served as the
reference point for displacement.
Each femur that survived the 2000 cycles was then loaded at 1 mm/sec under
displacement-controlled feedback until failure. Failure was defined as a
sudden drop of 445 N from the maximally observed
load9. Load and
displacement were recorded during testing, and the load versus displacement
curve was plotted. The load to failure (N) and stiffness (N/mm) were
calculated from these graphs (Fig.
4).
Differences in the mode of failure for the respective constructs were
observed during the testing process and were divided into two groups:
extra-articular and intra-articular failure.
Statistical Analysis
Within each group, a paired t test was used to compare differences between
treatments for load to failure and stiffness.
All but one femur from group II survived the initial 2000 cycles.
The femur that failed prematurely (1400 cycles) was reconstructed with
polymethylmethacrylate alone. Eight of ten group-III femora survived the 2000
cycles. The two femora that failed the cycling process were both reconstructed
with cement augmented with intramedullary Steinmann pins. All group-IV femora
survived the 2000 cycles.
The average load to failure for group-I femora reconstructed with cement
augmented with screws was 4732 ± 2287 N compared with 1877 ± 407
N for femora reconstructed with cement alone
(Table I). Within every matched
pair, the construct augmented with crossed screws was stronger (p = 0.025)
than the cement-alone construct.
In group II, the average load to failure was 7559 ± 1008 N for
specimens reconstructed with crossed screws and bone cement compared with 2363
± 898 N for femora reconstructed with cement alone
(Table I). Within every matched
pair, the construct augmented with crossed screws was stronger (p = 0.0007)
than the cement-alone construct.
The average load to failure for group-III femora was 4643 ± 2344 N
for those reconstructed with polymethylmethacrylate and crossed screws and
1525 ± 634 N for those reconstructed with intramedullary pins
(Table I). The construct
augmented with crossed screws was stronger (p = 0.019) than the intramedullary
pin augmented construct in every matched pair.
Group-IV femora confirmed that the host bone without any defect was
stronger (i.e., it had a higher load to failure) than each respective
contralateral femur reconstructed with crossed screws and cement (p = 0.0002)
(Table I). Four of the five
unmodified femora did not undergo failure at the maximal applied capacity
(8900 N) of the load-cell. The average load to failure for group-IV femora
reconstructed with cement augmented with screws was 5827 ± 506 N. The
one failure for the unmodified femur occurred at 7523 N.
Within each matched pair, the femora reconstructed with crossed screws had
significantly higher stiffness than the contralateral femora in group I (p =
0.0057), group II (p = 0.0024), and group III (p = 0.005)
(Table I).
The femora reconstructed with crossed screws failed through an
extra-articular fracture in seventeen of the twenty pairs
(Fig. 5). All fifteen femora
reconstructed with methods other than crossed screws failed through an
intra-articular fracture (Fig.
6, Table II).
Although there is much in the literature about cement reconstruction
of defects following curettage of benign aggressive tumors, this study is the
first, as far as we know, to demonstrate a technique that significantly
improves the strength of the reconstruction of noncontained defects of the
distal aspect of the femur. Screws were used because of the ease of insertion;
however, the configuration was based on a study in which Steinmann pins were
used to engage the intact
cortex13.
A load-to-failure biomechanical
analysis11 of
contained defects about the lateral condyle of the distal aspect of the femur
compared defects reconstructed with cement alone and defects reconstructed
with cement augmented with intramedullary Steinmann pins. In that study, the
addition of intramedullary pins to cement in contained defects did not provide
a clear biomechanical advantage. The authors noted that their results could
not be applied to pins placed in other configurations (i.e., through the
opposite cortex).
The addition of intramedullary Steinmann pins to polymethylmethacrylate for
the reconstruction of noncontained defects of the proximal part of the tibia
was investigated by Randall et
al.9. Defects were
created in the proximal-medial aspects of ten matched pairs of fresh-frozen
tibiae, and acrylic cement was press-fit around intramedullary pins that had
been placed to achieve an interference fit in the medullary canal. This
construct was then compared biomechanically with the contralateral tibia,
which was reconstructed identically but without the Steinmann pins. They also
evaluated the effects of large and small anchoring holes in the defects.
Augmentation of cement with intramedullary pins demonstrated a mechanical
advantage in noncontained defects, but it did not provide a significant
benefit if large anchoring holes were created in the defects bilaterally.
A more recent
analysis12 involved
contained defects in the proximal aspects of twenty-seven matched pairs of
formalin-fixed tibiae reconstructed with cement alone, cement augmented with
intramedullary Steinmann pins, or cement augmented with cortical Steinmann
pins. Each tibia was loaded at a constant rate of displacement, and the
resulting force-displacement curve was used to determine failure load and
stiffness. The constructs augmented with Steinmann pins (both cortical and
intramedullary) were statistically equivalent compared with the contralateral
tibiae reconstructed with cement alone.
We chose to examine load to failure and stiffness of reconstructed
noncontained defects in the medial femoral condyle. This model allowed better
evaluation of the true integrity of the replacement construct since there was
no host bone to provide substantial support and because the distal end of the
femur is a frequent site of giant-cell tumors. Applying load to the
reconstructed condyle alone instead of both condyles together was chosen to
better challenge the strength of the construct. The biomechanical testing
parameters for cycling and load to failure were similar to those used by
Randall et al.9. The
cycling parameters were established to simulate the loads across the medial
femoral condyle while walking with partial weight-bearing for one day.
In the present study, the crossed-screw construct was compared with a more
established technique, polymethylmethacrylate augmented with intramedullary
Steinmann pins. Screws were chosen because of the ease of application (i.e.,
no need to be cut or modified after placement). We believe that the hardware
configuration (rather than the hardware itself) provided the mechanical
advantage in the constructs. It is likely that similar results could have been
obtained with threaded Steinmann pins placed to engage the opposite
cortex.
The loading patterns across the knee joint for both normal and malaligned
knees have been examined and reported in the
literature14. For
normal subjects, the average joint loads across the knee were found to be 3.5
times body weight during single-leg stance of normal gait. Most of the load in
a normally aligned knee is localized to the medial compartment during walking.
A defect of the medial femoral condyle that has been reconstructed with
crossed screws is more likely to withstand the loads experienced during the
activities of daily living. For example, a 70-kg (686-N) adult with a normally
aligned knee joint would experience forces of approximately 2400 N at the
joint line. In this study, within each group, the average load to failure of
constructs without crossed screws was <2400 N. In contrast, the average
load to failure of constructs with crossed screws within each group was
>2400 N. The biomechanical advantage of the crossed-screw construct in this
study suggests that it may be clinically advantageous for noncontained defects
of the distal end of the femur.
Acrylic cement has been shown to undergo volume contraction of up to 5% on
curing15, and this
has been suggested as a factor in loosening at the bone-cement interface. In
the femora reconstructed with cement alone and with cement with Steinmann
pins, we observed gross motion of the cement mantle within the defect during
the cycling process. This motion was less in femora reconstructed with crossed
screws, as evidenced by their higher stiffness values, because the divergent
screws better constrained the cement in the irregular bone cavity.
The formalin-fixed femora (group I) may be considered to represent patients
with poor bone quality. This method of preservation has been demonstrated to
have no effect on maximal load capacity but to markedly diminish energy
absorption
capacity16. Since
it was believed that these femora would not survive cycling, this portion of
the testing was omitted for this group. In groups II and III, all femora that
were reconstructed with a combination of polymethylmethacrylate and crossed
screws survived the full number of cycles and failed at significantly higher
load to failure and stiffness than the contralateral reconstructed femora. The
unmodified femora within group IV had significantly higher load to failure and
stiffness than the contralateral femora reconstructed with crossed screws.
This comparison could not be rigorously quantified because four of the five
unmodified femora did not fail at the maximal capacity of the load-cell.
Examination of the data demonstrated a wide variability in the absolute values
of load to failure and stiffness for all groups. This difference is most
likely due to the variability of the femora themselves (e.g., bone quality and
size). To eliminate the effect of this variable, matched pairs were used so
that a more correct comparison could be made between the femora within each
pair.
The mechanism of failure was distinctly different in the femora with screw
augmentation than in those with other types of reconstruction. The screws
crossed the midline, allowing the cement mantle to remain relatively stable,
thus transferring most of the load to the bone proximal to the defect. A
transverse fracture through the lateral cortex and cancellous bone at the
level of the proximal bone-cement interface was observed at failure
(Fig. 5). An extra-articular
failure of this type can be treated with standard open reduction and internal
fixation. Conversely, femora reconstructed with cement alone and cement
augmented with intramedullary Steinmann pins did not have hardware that
crossed the midline. These femora failed through an intercondylar fracture,
which propagated proximally, displacing the cement and pins medially
(Fig. 6). This finding is
consistent with our clinical experience. Because the articular surface is
involved and the remaining bone stock is poor, failures of this type are
difficult to salvage with use of standard methods of fracture fixation;
osteoarticular allografting or endoprosthetic reconstruction frequently is
required.
While this study is conclusive for the specimens examined, there are
limitations that may be considered when designing future studies.
Biomechanical testing was conducted in only one loading configuration.
Additionally, the load application was not necessarily physiologic since the
intact lateral femoral condyle was not loaded. Future studies also could
incorporate loading contact that better simulates physiologic contact area and
location on the medial tibial plateau during a defined activity, such as gait.
Incorporating dual energy x-ray absorptiometry scanning into future protocols
would perhaps help to explain the wide variability in the absolute
load-to-failure and stiffness values among groups II, III, and IV. The effect
of cycling may have been more clear if data had been collected to determine
stiffness at the initiation of cycling. The average age of the donors of the
femora tested was higher than the average age of patients in whom most benign
aggressive bone tumors occur. However, the use of matched pairs helped to
minimize the variability of the study and allowed focus to remain on the
mechanical benefit of each construct. Benign aggressive bone tumors certainly
are not limited to the distal part of the femur, and the results from this
study cannot be extrapolated to other anatomical sites.
In this in vitro study, crossed-screw augmentation of
polymethylmethacrylate reconstruction of noncontained distal femoral defects
was mechanically superior to cement alone and to cement augmented with
intramedullary Steinmann pins. The screws can be placed easily under direct
vision, little additional operative time is required, and further dissection
is minimal. The strength and stiffness of this construct may allow earlier
rehabilitation and return to activity and may decrease the risk of
postoperative fracture. If a fracture occurs after this reconstruction, the
results of this study suggest that it is likely to be extra-articular and
easier to treat than the typical intercondylar fracture that occurs after the
other types of reconstruction examined. ?