Unicompartmental arthritis is traditionally thought of as a condition
affecting the femorotibial compartments, but it can also be isolated to the
patellofemoral articulation. In fact, isolated patellofemoral arthritis may
not be so
rare1-3.
In a radiographic study of patients over the age of forty years who had
painful knees, Davies et
al.2 noted that the
prevalence of isolated patellofemoral arthritis was 9% (nineteen of 206
knees). McAlindon et
al.4 performed a
radiographic study of 273 symptomatic knees in patients over the age of
fifty-five years and found a prevalence of isolated patellofemoral arthritis
of 8% in women. The condition also exists, albeit at a lower frequency, in
young and middle-aged people.
When encountering a patient with isolated patellofemoral arthritis for whom
a nonoperative approach and perhaps an arthroscopic débridement have
failed, the orthopaedic surgeon can choose from a number of operations,
including total knee arthroplasty and tibial tuberosity transfers as well as
lateral retinacular releases, patellofemoral replacements, autologous
chondrocyte implantation, facetectomy, removal of the patellar subchondral
bone, and denervation.
Although a total knee replacement can address any and all arthritic
conditions about the knee, the surgeon may consider this to be an overly
aggressive approach in the setting of unicompartmental arthritis. This would
be particularly true in a young, active patient, for whom the surgeon would
want the most bone-sparing procedure, and in an older, more frail patient, for
whom the surgeon would want to minimize surgical dissection, operating time,
and blood loss. The surgeon therefore can choose between a relatively extreme
procedure with predictable results (total knee replacement) and operations
demanding less surgical dissection and resection but offering less certainty.
In this review, we will summarize the current thinking regarding these various
options.
Pain is the primary symptom that should be addressed, and the surgeon must
first correctly identify the source of this pain. This can be particularly
challenging with the patellofemoral joint, as a large number of conditions can
refer pain to the anterior aspect of the knee. These conditions include
overuse, abnormal patellar tilt, plicae, neuromas, tendinitis, synovitis, and
focal lesions within the patella. In addition, pain can be referred from
elsewhere in the knee or from distant sites such as the hip or spine. It is
tempting to attribute pain to physical changes that are clearly visible on a
radiograph, on a magnetic resonance imaging scan, or to the naked eye;
however, the major source of a patient's pain may in fact be abnormal
intraosseous pressures, abnormal levels of biochemical mediators such as
proinflammatory cytokines or substance
P19, or other
factors that are not readily apparent. The enhanced ability of modern imaging
modalities to visualize the articular surface of the patellofemoral joint has
not improved the surgeon's ability to treat patellofemoral
pain20. This was
well stated by Insall when he noted "Curiously, neither the wide-spread
use of arthroscopy nor the advent of new diagnostic tests such as CT scanning
and magnetic resonance imaging have cast much light" on the enigma of
patellofemoral
pain21,22.
A patient with isolated patellofemoral arthritis typically describes
anterior knee pain when rising from a seated position and/or ascending
stairs23. The pain
is diminished when the subject walks on level ground. Pain at rest should
arouse suspicion of nerve-related pain, such as a neuroma, reflex sympathetic
dystrophy/complex regional pain syndrome, or a radiculopathy. More rarely,
pain at rest can be associated with a patellar tumor, an infection, or a
stress fracture.
On the physical examination, the surgeon should observe the patient in the
standing and walking position and look for a squinting (inward-pointing)
patella (Fig. 1), foot
pronation, and other signs of distant, pain-producing pathological conditions.
With the patient seated or supine, the hips should be evaluated for signs of
tightness, synovitis, and joint inflammation, since pathological conditions of
the hip can present as anterior knee pain. When assessing the knee, the
examiner should begin with light palpation of the soft tissues to detect a
neuroma or tendinitis. This is particularly important because no current
imaging study predictably reveals these conditions. If they are not diagnosed
at the time of the physical examination, they may be missed completely. The
presence or absence of patellar tilt needs to be determined on the physical
examination10,
since many imaging reports fail to make note of it.
A key sign of symptomatic patellofemoral arthritis on the physical
examination is tenderness of the lateral (or occasionally medial) facet of the
patella10. The
examiner assesses this by gently curling the fingers under the lateral (or
medial) border of the patella. In the setting of clinically meaningful
patellofemoral arthritis, this causes pain. (In applying pressure to the
facet, the examiner is simultaneously applying pressure to all of the soft
tissues between the skin and bone, including the retinaculum and the synovium.
The specific source of the pain can therefore be debated.)
The critical imaging study is the Merchant radiograph, for which the
patient is placed supine on the imaging table with the knees angled over the
end of the table and supported by a variable angle
device24
(Fig. 2). It is imperative that
there not be overlap between the patella and the underlying trochlea, so that
apparent approximation of the two structures represents true joint-space
narrowing rather than a radiographic artifact. Although the original article
by Merchant et al. described a knee flexion angle of
45°24, lesser
angles, such as 30°, are more desirable as they allow imaging of a more
proximal portion of the patellofemoral joint. Since trochlear dysplasias are
most commonly proximal, it is imperative that this portion of the compartment
be visualized. Although one may choose magnetic resonance imaging or
computerized tomography over a well-made Merchant radiograph, these are not
cost-effective methods for arriving at the same conclusions.
Having judged that the patellofemoral compartment is an important source of
pain, one must determine whether the arthritis is truly isolated to this
compartment. The absence of joint line tenderness is insufficient evidence
that femorotibial arthritis is absent, and one must not fail to order a
complete set of radiographs. These include not only standing anteroposterior,
true lateral25, and
Merchant radiographs, but also the standing tunnel view, also known as the
Rosenberg
view26,27
or the schuss view because of the crude resemblance to a schussing
skier23
(Fig. 3). When made adequately,
the standing tunnel view is indistinguishable from a supine tunnel view,
inasmuch as there is no overlap between the femur and tibia and the notch is
clearly visible. It is a critical projection because often it alone reveals
arthritis that is localized to the central and posterior aspects of the
femorotibial compartments (Figs. 4-A and
4-B).
Magnetic resonance imaging and arthroscopy can also be useful for
evaluating the femorotibial compartments, as can a nuclear bone scan, the only
imaging modality that provides a measure of a compartment's metabolic
activity.
A controversial issue is the extent to which nonarthritic chondral changes
(e.g., partial-thickness defects) can be present in other compartments before
the patellofemoral arthritis can no longer be considered isolated. Corpe and
Engh28 investigated
a similar issue in their evaluation of isolated unicompartmental arthritis of
the femorotibial articulation. In a study of patients who had undergone
femorotibial unicompartmental knee replacement, they noted that scattered
non-full-thickness chondral changes in the untreated compartments did not
affect the results. It is not known whether these findings can be extrapolated
to the patellofemoral joint.
Patellofemoral arthritis can be a subtle reflection of otherwise
subclinical inflammatory arthritis. Serum analysis for inflammatory conditions
such as rheumatoid arthritis is warranted, and in the United States an
infectious workup for Lyme disease should also be
considered29.
Lateral Retinacular Release
A lateral retinacular release should shift the patellofemoral contact area
medially. The operation can therefore be expected to be more successful in
patients in whom the medial articular cartilage is intact, but this has not
been studied, to our knowledge. It is also not known how far medial the
displacement must be for the operation to be successful. The procedure
incidentally denervates the patella to a certain extent, and it is unknown to
what degree that contributes to the success of the operation when it
works.
Aderinto and
Cobb41 reviewed the
results at an average of thirty-one months (range, twelve to sixty-five
months) after lateral retinacular release in fifty-three patients with
patellofemoral arthritis (with or without concomitant femorotibial arthritis).
Four patients required a knee replacement within eighteen months. Of the
remaining forty-nine patients, twenty-six had isolated patellofemoral
arthritis. Thirteen of those twenty-six patients were very satisfied or
satisfied with the result of the lateral release, whereas the other thirteen
were dissatisfied. Unfortunately, the authors did not differentiate between
patients with lateral arthritis and those with global (patellar and trochlear)
involvement.
Total Knee Replacement
A traditional total knee replacement, including patellar resurfacing,
removes all present and future sources of arthritis. Although early failures
continue to be
reported42, the
longevity of many knee-replacement designs is excellent and may be
substantially greater than the ten to fifteen years that have been reported in
the past43.
In a study of fifty-three patients who had been followed for an average of
7.4 years after total knee replacement with resurfacing of the patella to
treat isolated patellofemoral arthritis, Laskin and van
Steijn44 noted an
average range of motion of 122°, compared with 117° in a separate
cohort of patients with tricompartmental disease, with 81% (forty-three) of
the fifty-three patients having a good-to-excellent result.
Parvizi et al.45
evaluated the results of thirty-one knee replacements in twenty-four patients
at an average of five years. They noted that twenty-one patients required a
lateral retinacular release and three more required a more extensive
realignment.
In a study of the results of thirty-three replacements in twenty-seven
patients who had primarily patellofemoral arthritis, Mont et
al.46 reported
twenty-eight excellent results, one good result, and one poor result at a mean
of eighty-one months postoperatively. The mean Knee Society objective score
increased from 50 points preoperatively to 93 points postoperatively.
Resurfacing of the patella in total knee replacement is a controversial
issue, even when the patella is arthritic. A number of surgeons have reported
routinely leaving the patella unresurfaced during knee replacement surgery.
Thompson et al. took the concept to the extreme when they performed
thirty-three total knee replacements without patellar resurfacing in
thirty-one patients with isolated patellofemoral arthritis in whom the pain,
by definition, could have come only from the patellofemoral
joint47. At twenty
months postoperatively, twenty-one knees were pain-free and twelve knees were
occasionally painful. The average range of flexion was 104°. None of the
patients required revision surgery.
Despite the success of total knee arthroplasty, many surgeons consider it
to be too big a sacrifice of healthy tissue and too great a surgical
dissection for a patient with disease involving mainly one compartment.
Anterior Transfer of the Tibial Tuberosity
If patellofemoral pain is postulated to be the result of loads applied to
deficient cartilage, it stands to reason that a diminution of those loads
could lead to pain relief. This has been the theory behind operations that
relocate the tibial tuberosity to a more anterior position. However, the size
of the contact area on which these loads are applied has to be considered when
judging subchondral stresses. Indeed, as noted by Lewallen et
al.48, a diminution
of forces may not necessarily lead to a reduction of the stresses. Therefore,
studies that only evaluate
forces49 may not
provide clinically relevant data.
Maquet50 was the
first to develop a procedure that moves the tibial tuberosity anteriorly, with
his eponymous operation that elevates the tuberosity by 2.5 cm. On the basis
of in vitro studies, Ferguson et
al.51 later
proposed a variation on Maquet's operation, whereby the tibial tuberosity was
elevated just 0.5 in (1.27 cm). Schepsis et
al.52 recommended
an elevation of 1.6 to 1.8 cm to compensate for the expected settling of the
construct. Subsequent investigators have recommended elevations as low as 1.25
cm53 and even 1
cm54,55
on the basis of in vitro considerations. Ferrandez et
al.55 noted that
elevation of the tibial tuberosity can paradoxically increase stresses on the
proximal portion of the patella, particularly when it is elevated >1 cm.
Using in vitro measurements, both Nakamura et
al.54 and Ferrandez
et al. found an elevation of >1 cm to be counterproductive with regard to
patellofemoral stresses.
The Maquet operation has been reportedly associated with a number of
complications, most notably skin necrosis, cracking of the osteotomy fragment,
patellar tendinitis, and the development of a painful prominence at the level
of the tibial
tuberosity52. These
problems have largely been addressed by lengthening of the osteotomy fragment
that is elevated, as this minimizes the angulation required for a given
elevation.
Because Maquet's operation has been used for a wide variety of
patellofemoral conditions and because the results have not always been
correlated with the etiology of the pain or the specific location of the
arthritis, the clinical results of the operation for patellofemoral arthritis
can be difficult to ascertain. Heatley et
al.56 noted that,
of fourteen knees treated with a Maquet operation for patellofemoral
arthritis, six were rated as good to excellent, three were rated as fair, and
five were rated poor at six years postoperatively. Jenny et
al.57 evaluated
sixty-five patients who had undergone a Maquet procedure for a variety of
conditions and found that the forty-eight patients with patellofemoral
arthritis had the best results at an average of eleven years. In neither study
was the exact location of the arthritis within the patellofemoral joint
specified. Division of the patients into subgroups according to the location
of the chondral lesions would have been useful.
Engebretsen et
al.58 reported on
thirty-eight patients treated with the Maquet procedure. Of thirty-three
patients who were available for follow-up at a mean of five years, ten had
improvement, seventeen had no change, and six were worse off. The patients
with improvement were noted to have "Grade III and IV" cartilage
changes primarily involving the lateral facet.
Over the last twenty years, Maquet's operation has diminished in popularity
in Europe59, with
some surgeons actually reversing the
procedure23.
However, a variation on the Maquet procedure—anteromedial displacement
of the tibial tuberosity—increased in popularity in the United States
over that same period of time.
Anteromedial Transfer of the Tibial Tuberosity
The anteromedial tibial tuberosity osteotomy, as described by Fulkerson et
al.60, is a
modification of the Maquet operation. The procedure is carried out with an
oblique osteotomy of the tibial tuberosity that is sloped in an
anteromedial-to-posterolateral direction. The tibial tuberosity is transferred
both medially and anteriorly, with the relative proportion of each transfer
being determined by the slope of the cut. A relatively flat, horizontal cut
across the tibial tuberosity leads to mostly medial displacement, whereas a
more vertical cut leads to greater vertical displacement. A 45° cut leads
to equal medial and vertical displacements.
This operation has the disadvantage of not allowing the same amount of
anterior displacement as Maquet's procedure permits, and it displaces the
tibial tuberosity medially even in patients in whom the need for such
medialization is unclear (for example, patients with a normal quadriceps [Q]
angle)61. Medial
displacement of the tuberosity in this setting may increase the medial
femorotibial forces, resulting in pain in that compartment.
This procedure does offer a number of advantages. A single osseous cut
achieves both anterior and medial displacement in patients who need this
combined effect. A bone wedge is not required at the osteotomy site to
maintain the displacement. If the screws used for fixation of the shingle are
oriented perpendicular to the plane of the osteotomy, both the drill and the
screws will be oriented toward the medial aspect of the posterior part of the
tibia and the popliteal space, where injury to the popliteal artery or vein is
unlikely62. Most
importantly, in the setting of patellofemoral arthritis this operation shifts
the patellofemoral contact area medially, an advantageous displacement when
the arthritis is located on the lateral aspect of the patella.
This osteotomy presents technical pitfalls. It requires two cuts. The first
cut involves all but the most proximal portion of the tibial tuberosity. That
cut is angled in an anteromedial-to-posterolateral direction, with care taken
to protect both the anterior tibial artery and the deep peroneal nerve
posteriorly. The second cut begins at the most proximal and lateral aspect of
the first cut and is oriented toward the most proximal portion of the tibial
tuberosity. If the surgeon fails to carry out this second cut and chooses
instead to carry the first cut all of the way to the top of the tuberosity,
the cut will involve a large part of the proximal tibial metaphysis.
If the tibial tuberosity were displaced 11 mm along a 45° degree plane,
which is a substantial displacement, this would lead to 8 mm of medial
displacement and 8 mm of anterior displacement. Ateshian and
Hung7 calculated
that this would result in only a 10% reduction in stress. This would explain
why the procedure is not well suited for the treatment of global arthritis of
the patellofemoral
joint63, although
the correlation between clinical success and stress reduction has yet to be
established. Moreover, an 8-mm anteromedial displacement of the tibial
tuberosity would lead to a 4.5-mm medial displacement of the patellofemoral
contact area7,
which, in some cases, may be enough to provide clinical improvement. Greater
anterior advancement is not necessarily beneficial. For example, Takeuchi et
al.64 performed an
oblique osteotomy (with a 70 × 15-mm shingle) on seven specimens, and,
using pressure-sensitive film, found that an elevation of 10 mm provided
greater stress (force/area) reduction than an elevation of 20 mm. Indeed, the
diminished forces at elevations of 20 mm were largely (and negatively) offset
by the diminution in contact area.
Pidoriano et
al.63 correlated
the results of this procedure with the specific location of the arthritic
lesions. They noted the best results in patients with arthritis isolated to
the lateral portion of the patella, even when there was a complete loss of
articular cartilage at that site, and the worst results in patients who had
diffuse involvement of the patella, extensive involvement of the trochlea,
diseased cartilage at the proximal portion of the patella, and crush
injuries65.
Complications associated with this operation include postoperative fracture
of the tibial shaft and nonunion of the osteotomy site. The osteotomy
incorporates a substantial portion of the proximal tibial cortex, thus greatly
diminishing the tibia's ability to withstand torque. Stetson et
al.66 reported that
six patients and Bellemans et
al.67 reported that
four patients sustained a fracture of the tibia at the distal junction of the
osteotomized tibial tuberosity and the tibial shaft. All fractures occurred as
the patients progressed to full weight-bearing status. Bellemans et al.
recommended that patients remain nonweight-bearing for eight weeks. In the
series reported by Stetson et al., two of the six fractures occurred after
eight weeks, and the authors therefore recommended that full weight-bearing
not be allowed until there was radiographic evidence of osseous healing.
The screw heads may be palpable and tender, and one or both screws may need
to be removed after healing of the osteotomy site. Removing a partially
threaded cancellous screw can be difficult as the threads may not be able to
cut back through bone that has grown along the nonthreaded portion of the
screw, and use of a fully threaded screw may be advisable.
Patellofemoral Replacement
The practice of resurfacing the patella in patients who have patellofemoral
arthritis predates total knee replacement by approximately ten years.
McKeever68 was, to
our knowledge, the first to report such an operation when he described fixing
a metallic implant to the undersurface of the patella by way of a transverse
screw. Vermeulen et
al.69 reported
that, eight to ten years following use of the McKeever prosthesis in nine
women (age range, forty-six to seventy years; average age, sixty-one years),
no patient had required a reoperation; pain and function scores were not
reported. In 1979, Pickett and
Stoll70 reported
that, one to twenty-two years following forty-six McKeever operations,
thirty-nine had a "satisfactory" result as determined by both the
patient and the surgeon.
Harrington71 found
that, at five years after McKeever patellar resurfacing, seventeen of
twenty-four patients had a good or excellent result. The poor results occurred
in patients who had signs of osteoarthritis in other parts of the knee.
In 1975, Aglietti et
al.72 introduced a
dome-shaped cobalt-chromium-molybdenum patellar component that was identical
in shape to the patellar button of their total knee replacement. They did not
report the results of clinical follow-up. Worrell designed a cobalt-chromium
device and reported short-term follow-up results in
197973 and then
again reported results in
198674. Of fourteen
knees in thirteen patients (eighteen to thirty-eight years of age) followed
for a mean of five and a half years (range, one to eight years), two had an
excellent result; one, a good result; seven, a fair result; and four, a poor
result. The best results were in patients over the age of forty years who had
"extremely poor quadriceps function." The implant has not been
reported on by other investigators, to our knowledge.
The concept of a metallic implant articulating with the cartilaginous
trochlea has been unappealing to the orthopaedic community because of concerns
about harming an intact trochlea or, conversely, not addressing chondral
lesions already present in the trochlea.
In 1979,
Lubinus75 and
Blazina et al.76
introduced separately the concept of a patellofemoral replacement whereby both
the patella and the trochlea were resurfaced. Despite some confusion in the
literature46, this
operation is separate and distinct from a simple patellar resurfacing. The
procedure requires less surgical dissection than a total knee replacement,
removes less bone, preserves the femorotibial compartments as well as the
cruciate ligaments, does not necessitate blood transfusion, and can
potentially be less expensive.
Patellofemoral replacement surgery has not been widely accepted. The
original descriptions of the procedure did not include strict criteria,
technical pitfalls were not yet appreciated, and little emphasis was placed on
realignment of the extensor mechanism of the knee. Consequently, reports
pertaining to the earliest designs showed disappointing results. In Europe
this led to efforts to redesign the implant and improve the operative
technique, and in the United States most surgeons simply abandoned the
procedure.
The literature of the last fifteen years has been far more encouraging.
Cartier et al.77
reported the results at an average of four years after patellofemoral
replacements in seventy-two patients, thirty-six of whom had a concomitant
femorotibial unicompartmental replacement and one of whom also underwent a
tibial osteotomy. Sixty-one patients (85%) had a good-to-excellent result
according to the Mansat (0 to 20-point)
scale78. Another
five patients (7%) had a good-to-excellent result after undergoing a remedial
procedure (additional patellar realignment and exchange of the patellar button
for a smaller size). In 1995, Argenson et al. reported the results of
sixty-six replacements reviewed at an average of 5.5
years79. Dividing
their results by etiology, they noted a good result in twenty of twenty-two
patients with malalignment and dysplasia, nineteen of twenty patients with
posttraumatic arthritis, and seventeen of twenty-four patients with arthritis
of unknown etiology.
In 1996, Krajca-Radcliffe and
Coker80 reported a
good-to-excellent result in fifteen of sixteen patients followed for an
average of 5.8 years after patellofemoral replacement. Mertl et
al.81 reviewed the
outcomes of fifty patellofemoral replacements at an average of three years
(with twenty-two followed for at least four years) and, using the Guepar
rating tool82,
found 34% (seventeen) very good, 48% (twenty-four) good, and 18% (nine) poor
results. Of the nine failures, five pertained to the femorotibial
compartments; two, to a complex regional pain syndrome; one, to an infection;
and one was in a patient with a Workers' Compensation claim whose pain
remained unresolved. Arnbjornsson and
Ryd83 found that,
at seven years after patellofemoral replacement, 75% (eighty-five) of 113
patients were satisfied with the result, 58% (sixty-six) of the 113 walked
without assistive devices, and 44% (fifty) of the 113 had no or only
occasional knee pain. Predictably, they found the poorest results in patients
who had the nonspecific diagnosis of chondromalacia.
De Cloedt et
al.84 followed
forty-five patients for three to twelve years after patellofemoral replacement
and found that only 43% (nine) of twenty-one patients with arthritis and no
malalignment or dysplasia had a good result, with the main cause of failure
being degeneration of the femorotibial compartments. In contrast, 83% (twenty)
of twenty-four patients with so-called patellofemoral instability and/or
trochlear dysplasia had a good result. Kooijman et
al.85 followed
forty-five patients for ten to twenty-one years after patellofemoral
arthroplasty. Fifteen patients required another operation (most commonly a
soft-tissue operation such as a lateral retinacular release) soon after the
index procedure, and twelve underwent either a tibial osteotomy or a total
knee replacement at an average of fifteen years. However, two-thirds
(thirty-three) of the forty-five patients still had the patellofemoral
replacement at the time of final follow-up (at an average of seventeen
years).
The poor results reported after patellofemoral replacement have a common
theme: failure to appreciate present or incipient femorotibial arthritis.
Arthritis is more likely to develop in the remainder of the knee in a patient
with an incomplete evaluation and in a patient in whom the patellofemoral
arthritis does not have a clear origin, such as malalignment, dysplasia, or
trauma.
Autologous Chondrocyte Implantation
Autologous chondrocyte implantation involves harvest of articular cartilage
from a portion of the patient's femoral condyle, culture of the chondrocytes
to multiply their number, and reimplantation of the chondrocytes into the
chondral defect(s) that is (are) thought to be the source of pain. Although
this technology has been used mainly for isolated defects of the femoral
condyle, attempts have been made to apply it to the patellofemoral joint.
Early attempts in a small number of patients did not yield good
results86. This was
thought to be related to the unfavorable mechanics of the patellofemoral
joint87,88.
Indeed, both sides of the patellofemoral articulation are subject to shear
stresses with every flexion-extension cycle. The central portion of the
patella is twice subject to shear during the cycle: at a knee flexion angle of
approximately 45° and again at
120°12.
Recently, Minas and
Bryant89 reported
more favorable results with autologous chondrocyte implantation. They
investigated a group of forty-five patients who had patellofemoral arthritis,
with or without femorotibial arthritis. Eight patients had isolated patellar
arthritis, nine had trochlear arthritis, four demonstrated patellar and
trochlear arthritis, and the remainder exhibited arthritis in two or more
compartments. The authors noted that, at an average of two years
postoperatively, the patients with patellar arthritis, trochlear arthritis,
and "patella plus trochlea plus weight-bearing condyles"
(involvement of both the patellar and the trochlear side of the patellofemoral
articulation as well as of the weight-bearing portion of one or both femoral
condyles) "all had marked improvement in pain relief and
function." There were eight graft failures in the patellofemoral
compartment. Minas and Bryant noted clinical failures in five of eleven
patients with a Workers' Compensation claim.
Autologous chondrocyte implantation is expensive but may eventually be
cost-effective if it provides long-term pain relief. Examining costs per
quality-adjusted life-year, Clar et
al.90 attempted to
compare the cost-effectiveness of autologous chondrocyte implantation
procedures in the knee with microfracture and mosaicplasty operations, but
they found the results to be inconclusive because of a lack of sufficient
long-term data. Derrett et
al.91 also
investigated the costs per quality-adjusted life-year of both autologous
chondrocyte implantation and mosaicplasty. They found that both operations
fell below "an implicit English funding threshold"—in other
words, they were not (yet) cost-effective according to the available follow-up
data.
Longer follow-up and clinical trials comparing autologous chondrocyte
implantation with other options for treatment of patellofemoral arthritis are
needed before this option can be recommended as a standard procedure.
Patellectomy
A patellectomy is a resection arthroplasty of the patellofemoral joint. The
term patellectomy encompasses a number of operations, the end result
of which is the removal of the patella. These operations differ only in the
way that the peripatellar soft tissues are managed. Patellectomy was a common
operation in the early to mid-twentieth century, when a number of structures
about the knee, including the patella and menisci, were thought to be
expendable. Its appeal lies in its perceived relative simplicity compared with
other procedures. It can be reasoned that removal of the patellar half of a
painful patellofemoral joint will cause the pain to disappear. However, the
pain from an arthritic trochlea cannot be relieved by a patellectomy, and
removal of the patella greatly diminishes the lever arm of the extensor
mechanism of the knee. This can lead to extensor weakness and/or a so-called
extensor lag, whereby the patient is incapable of completely straightening the
knee.
The popularity of the procedure has gone through cycles. In 1909,
Heineck92 condemned
it, stating that, on the basis of his experience with five patients,
"its removal is invariably followed by impairment of power, by some
functional loss."
Brooke93, in 1937,
reported good results in thirty cases, and Hey
Groves94 supported
Brooke's opinion that the procedure was beneficial. In 1948,
McFarland95 went so
far as to recommend the procedure for simple recurrent patellar dislocation.
However, in 1949,
Scott96 reviewed a
series of seventy-one patellar fractures treated with patellectomy and found
that 60% of patients reported so-called giving-way of the knee and 90% of
patients still had pain. Since then, the literature has been inconclusive
regarding this procedure.
In 1969, Castaing et
al.97 reported that
forty-six of sixty-one patients had a good-to-excellent result at an average
of five years after a patellectomy. The patellectomy was carried out with use
of a vertical incision over the patella. A nearly full-thickness flap of
quadriceps tendon was turned down and was sutured into the patellar tendon,
and the medial and lateral retinacula were sutured over this turn-down
construct. In a subset of seventeen patients with arthritis, just six patients
had a good-to-excellent result, with the poorest results found in those with
femorotibial disease. Overall, quadriceps strength, the active range of
motion, and knee stability were judged to be acceptable. Stair descent tended
to be a source of problems. Heterotopic ossification about the patellectomy
site (considered to be "patellar regeneration" by the authors) was
noted in two patients, but it had no correlation with results. Recovery tended
to be long, and a number of patients noted continued improvement over the
follow-up period.
Compere et al.98
reported on twenty-nine knees followed for an average of approximately seven
years after patellectomy. The patients were young (average age, 43.5 years)
and presented with various degrees of arthritis. The fibers at the dorsum of
the patella were maintained while the patella was everted and enucleated. The
medial and lateral borders of the quadriceps and patellar tendons were sutured
together to fashion a tube. The vastus medialis was advanced and sutured onto
this tube. Ninety percent (twenty-six) of the twenty-nine knees were rated as
having a good-to-excellent result. "Symptomatic calcification in the
patellar tendon" developed in one patient; it was surgically removed,
and the patient eventually had an excellent result. De la
Caffinière99
noted good results in a cohort of seventy young patients (thirty to sixty
years of age) with patellofemoral arthritis. Addressing the issue of
quadriceps weakness and extensor lag following patellectomy procedures, he
noted that final quadriceps strength could not be assessed less than two years
following the operation. Baker and
Hughston100 noted
that nineteen of twenty patients with arthritis of the knee were satisfied at
an average of fourteen years following treatment with the Miyakawa
technique101,
which requires that a partial-thickness strip of quadriceps tendon be turned
down over the void left by the patella. In addition, the vastus lateralis and
vastus medialis are advanced, crossed over each other, and attached to the
quadriceps tendon.
In a more recent review, Lennox et
al.102 found the
poorest results in their patients who underwent a patellectomy for arthritis.
(Not all of their patients underwent the patellectomy for that indication.)
They reported a good result in only 54% (twelve) of twenty-two patients with
arthritis, with 27% (six) of the twenty-two patients reporting that they felt
worse than they had preoperatively. This was a retrospective analysis of
patients who had been operated on twelve to forty-eight years earlier, with
the charts having been culled from hospital records. The evaluation was
carried out mostly by telephone. The article does not name or describe the
surgical technique(s).
A shortcoming of all of the above studies is a lack of description of the
anatomic location of the arthritis. It is not clear, for example, what
percentage of patients had arthritis limited to the patella or how often
arthritis was present on both the patella and the trochlea.
We believe that a patellectomy can relieve symptoms and improve function
when the advancement of soft tissues adequately compensates for the void left
by the patellectomy. Although logic would dictate that patellectomies
performed for the treatment of patellofemoral arthritis would be more
successful when the arthritis is limited to the patella, this has not been
studied, to our knowledge. A major drawback of a patellectomy is that, for all
practical purposes, it is irreversible, although perhaps less so than other
resection arthroplasties.
Removal of Subchondral Bone, Facetectomy, Thinning, and
Denervation
These infrequently performed procedures approach the problem of
patellofemoral arthritis in a manner that, in the twenty-first century, is
considered unconventional.
In a procedure termed spongialization by its originators (Ficat et
al.103), the
subchondral bone is removed down to cancellous (spongy) bone. The operation
was also described by
Marmor104 and is
still used by some in lieu of patellar resurfacing during total knee
replacement
arthroplasty105.
When the patellofemoral arthritis is limited to the lateralmost portion of
the compartment, consideration can be given to removing 1 to 1.5 cm from the
lateral aspect of the patella (a so-called facetectomy). Yercan et
al.106 followed
eleven patients for an average of eight years after such a procedure and noted
a significant increase in the average pain score (p = 0.04) and in the average
Knee Society score (p = 0.02) (based on the findings of a physical
examination)107.
The average Knee Society functional score (based on walking and
stair-climbing) also improved.
In order to diminish the stresses on the patellofemoral joint, bone can be
removed from the center of the patella to decrease its thickness. Both the
removal of the layer and the reattachment of the two remaining portions may be
technically difficult, especially since care must be taken not to further
damage the articular surface. Nerubay and
Katnelson108
performed the procedure on fifteen patients who had demonstrated patellar
malalignment and followed them for an average of three years. Twelve patients
had a good-to-excellent result. Vaquero and
Arriaza109 used a
double saw to remove 7 mm of bone from the center of the patella and noted a
diminution of patellofemoral stresses when they applied Fuji film to the
patellofemoral joint. Following this operative procedure, there may not be
much osseous bed left for a prosthetic replacement if one is needed at a later
time.
The patella is innervated by multiple superficial sensory nerves, including
the medial cutaneous nerve of the thigh, the lateral femoral cutaneous nerve,
the medial and lateral retinacular nerve, and the anterior femoral cutaneous
nerve110. Division
of these nerves can diminish pain emanating from the patella but is likely to
be less effective for trochlear lesions. No formal study of surgical
denervation has been carried out in the specific setting of patellofemoral
arthritis, to our knowledge.
A number of challenges remain in the field of patellofemoral arthritis. In
order to improve investigations of treatments, the orthopaedic community needs
to accept a validated outcome tool. The term patellofemoral arthritis
itself requires greater anatomic precision, as arthritis localized to one
portion of the patellofemoral compartment may behave differently from global
patellar and trochlear arthritis. Clinically, the challenges to the surgeon
remain the identification of patients whose symptoms are truly the result of
isolated patellofemoral arthritis and the selection of a viable surgical
operation for those patients. The lack of uniform reporting makes it difficult
to recommend one procedure over another at this time. Operations that shift
the patellofemoral contact area medially are recommended for young patients
with arthritis that is limited to the lateral aspect of the patella, whereas
joint replacement surgery remains the best option for older patients who have
diffuse arthritis involving both the patella and the trochlea.