Demographic information and intraoperative data were gathered,
as part of a prospective database, on all patients (600 knees) undergoing
primary total knee arthroplasty at the senior authors' (J.J.J. and A.G.R.)
institution since the mid-1980s. The present study included patients
in whom primary total knee arthroplasty was performed by the authors
between June 1991 and June 1997 and who sustained an intraoperative
injury to the medial collateral ligament as documented in the database
and confirmed by evaluation of the operative report. In order to
minimize the likelihood that the injury took place in an already
damaged, attenuated, or insufficient ligament, patients with any
measurable valgus on a preoperative lower-extremity mechanical axis radiograph
(or with a preoperative anatomic axis of more than 7° of valgus)
were excluded. A total of sixteen knees (2.7%) in fourteen patients
met these criteria and were included in the study. Review of the
operative reports confirmed that the injuries were indeed midsubstance
transections that were repaired or avulsions off the tibia or femur
that were reattached to bone, and not simply extensive releases
as part of an attempt to gain collateral ligament balance in tight
varus knees.
Preoperatively, demographic and clinical data, including age,
weight, and height, were obtained for all patients, and The Hospital
for Special Surgery knee score19 and
the body-mass index were calculated for all patients as well. A
patient with a body-mass index of greater than 30 is considered
obese, and one with an index of greater than 40 is considered morbidly obese5. Preoperative weight-bearing radiographs
of all of the affected knees were available for review. They were
used to calculate the preoperative anatomic axis of the lower extremity
by measurement of the angle subtended by lines drawn down the femoral and
tibial medullary canals. In addition, full-length hip-to-ankle mechanical
axis radiographs of twelve of the sixteen knees (eleven of the fourteen patients)
were available for review.
The surgical techniques employed by the two surgeons were similar
throughout the period of study. All operations were performed by,
or under the direct supervision of, one of the two senior authors. A
midline straight longitudinal skin incision was used in all but
two knees; in those two knees, a curved medial incision was made
to incorporate a previous incision. A standard medial parapatellar approach
was employed for exposure in all knees.
Following eversion of the patella, flexion of the knee, and transection
of the anterior cruciate ligament, a scalpel or electrocautery was
used to find a subperiosteal plane beneath the deep medial collateral
ligament, beginning at the tibial joint line and proceeding distally.
This medial sleeve was raised off the tibia, with an electrocautery
and an osteotome moving from anterior to posterior under direct
vision. As the dissection proceeded posteriorly, less was taken
distally in order to preserve the insertion of the medial collateral
ligament on the tibia. In knees that were in neutral or minimal
varus alignment, dissection proceeded to the midcoronal plane at
the level of the joint line. In knees with more severe varus deformity,
this subperiosteal elevation was carried around to the posteromedial
corner of the knee, with the knee in flexion and the tibia gently
drawn forward. Attempts were made, with use of judiciously placed
retractors, to protect the sleeve of the medial collateral ligament
from injury during tibial and femoral preparation.
Intramedullary instrumentation was used on the femur and extramedullary
instrumentation was used on the tibia in all patients. Injury to
the medial collateral ligament was typically identified during assessment
of ligament balance with the trial components; uncommonly, it was
noted after the final implants were cemented into place.
When injury of the medial collateral ligament was identified,
the location of the injury (the femoral origin, tibial insertion,
or midsubstance) was determined and repair was performed. The ligament
was exposed, and avulsions off bone were repaired with nonabsorbable
suture anchors (Mitek [Ethicon, Somerville, New Jersey] or Statak
[Zimmer, Warsaw, Indiana]) of nonabsorbable, heavy, braided suture
material. A screw-and-washer reattachment was performed in one knee
with an avulsion of the ligament from the distal part of the femur.
Intrasubstance ligament lacerations were repaired with interrupted,
nonabsorbable, heavy, braided sutures. The thickness of the tibial
polyethylene spacer used in each patient was determined after the
ligament had been repaired, with the goal of balancing the knee
in flexion and extension as would be done in any knee arthroplasty.
Postoperatively, all patients were permitted unrestricted weight-bearing
while wearing a hinged knee brace that allowed a free range of motion. Apart
from the full-time use of the hinged knee brace, aftercare was not
different from that of patients who had not sustained an intraoperative injury
of the medial collateral ligament. It included twice-daily physical
therapy that emphasized range of motion and gait while the patient
was in the hospital, and home physical therapy or inpatient rehabilitation
following discharge. The patients did not use a continuous-passive-motion
machine.
During the period of study, we preferred to use unconstrained
total knee implants (NexGen or Miller-Galante-II components; Zimmer)
whenever possible; these implants included a flat tibial polyethylene
bearing and provided for retention of the posterior cruciate ligament.
When a severe varus or severe flexion-varus deformity was present,
a posterior stabilized implant with minimal constraint (NexGen Legacy
Posterior Stabilized or Insall-Burstein-II Posterior Stabilized
components; Zimmer) was used. In the present series, twelve knees had
unconstrained implants and four had posterior-stabilized implants.
Postoperatively, patients were encouraged to return to the physician's
office for clinical and radiographic follow-up. At the latest follow-up
evaluation, the range of motion was measured, The Hospital for Special
Surgery knee scores were calculated, and weight-bearing radiographs
were reviewed. The anatomic axis was calculated from the weight-bearing
radiographs at the last follow-up evaluation, to determine whether
an acceptable overall alignment had been restored to the joint. Patients
were queried about their subjective sense of joint stability and
about any need to use a knee brace for walking. On physical examination,
stability of the medial collateral ligament was assessed by manually
applying a valgus stress to the knee at both 0° and 30° of knee
flexion. Opening of more than 5 mm in either position, and the absence
of a firm end point to valgus stress in either position, were used
as the criteria for defining failure. In all cases, the physical
examination, review of the radiographs, and determination of The
Hospital for Special Surgery knee scores were performed by orthopaedic
surgeons other than the operating surgeons.
Sixteen knees in fourteen patients were available for follow-up
at a mean of forty-five months (range, twenty-four to ninety-five
months). The mean age at the time of surgery was sixty-three years
(range, forty-seven to eighty-six years). Four of the patients were
male and ten were female. The preoperative diagnosis was osteoarthritis
in fifteen knees and rheumatoid arthritis in one; one patient (one
knee) had a history of high tibial osteotomy. The mean body-mass
index of the patients in this series was 32.5 (range, 20 to 49);
according to defined criteria5,
three patients were classified as morbidly obese; three, as obese;
and eight, as normal with regard to weight. No patient was lost
to follow-up, and twelve patients (fourteen knees) returned for
in-person follow-up, including complete clinical and radiographic
evaluation. Two patients (two knees) were unable to return for physical
examination. One of the two was examined and had radiographs made
by her local physician, and these were sent to us for review. The
other patient was unable to return for an in-person visit because
of medical conditions and family-related issues. That patient responded
to a telephone questionnaire that was administered for the purpose
of determining a knee score as well as the patient's subjective
sense of knee stability and function, which-although good-may have
been influenced by her activity level.
No patient in the series required bracing beyond six weeks postoperatively,
and no patient reported subjective instability of the knee. No patient
had demonstrable laxity in the coronal plane on physical examination
either in 30° of flexion or in full extension. All patients were
able to walk in the community at the time of the last follow-up.
No one underwent revision arthroplasty.
Preoperatively, the mean Hospital for Special Surgery knee score
was 47 points (range, 31 to 77 points); at the time of final follow-up,
the mean score was 93 points (range, 78 to 100 points). Thirteen
knees had an excellent result (a score of 85 points or higher),
and three had a good result (a score between 70 and 84 points).
There were no fair or poor results. Active extension of the knee improved
significantly, from a mean of 6° preoperatively to a mean of 2°
at the time of final follow-up (p = 0.011). With the numbers available,
there was no significant difference between preoperative flexion
and flexion at the time of final follow-up (106° compared with 108°,
p = 0.53). Fifteen of the sixteen knees achieved flexion of at least
90° (maximum, 125°); the other knee could flex 85° actively and
had a final Hospital for Special Surgery score of 78 points (the
lowest in the series). In one other patient, a 15° flexion contracture
developed following arthrotomy, débridement, and exchange of the polyethylene
bearing, with retention of all of the cemented components, for the
treatment of an acute hematogenous infection that developed twenty-two months
following the total knee arthroplasty. At forty months following
the index arthroplasty (eighteen months following the arthrotomy
and débridement), the patient remained free of infection without
suppressive antibiotics. That patient also required manipulation
under anesthesia following the index arthroplasty, and that manipulation
was complicated by a transient, partial peroneal nerve palsy, which
fully resolved. Of note, that patient had a 35° flexion contracture
and a 20° varus deformity preoperatively. There were no other reoperations
in this series.
All of the injuries of the medial collateral ligament were believed
to be the result of avulsion by medially placed retractors or the
result of direct injury from either the oscillating saw-blade during
the tibial cut or a sharp instrument used for medial subperiosteal
elevation. Four of the injuries of the medial collateral ligament
were avulsions; three of these avulsions were off the tibial insertion,
and one was off the femoral origin in a severely osteopenic patient.
Twelve of the injuries were midsubstance disruptions of the ligament.
Radiographic assessment demonstrated a preoperative anatomic
axis that averaged 7.1° of varus (range, 20° of varus to 5° of valgus).
Postoperatively, the mean anatomic axis was 5.9° of valgus (range,
neutral to 10° of valgus); this difference was significant (p < 0.01).
At the time of final follow-up, two knees displayed small (less
than 1-cm), nonprogressive radiolucencies under the medial portion
of the tibial component. No radiolucencies were observed in association
with the bone anchors or screws used to reattach the medial collateral
ligament. There were no clinical complications associated with hardware
used to reattach the medial collateral ligament.
Disruption of the medial collateral ligament during primary total
knee arthroplasty is a serious complication. Since coronal-plane
instability can result in the need for revision, nearly every report
of which we are aware has advocated using unlinked constrained or
hinged implants in patients with insufficiency of the medial collateral
ligament that either is noted preoperatively or is the result of
iatrogenic injury1-4,11,20,21.
To our knowledge, the present study, which documented a 2.7% rate
of inadvertent injury of the medial collateral ligament (sixteen
of 600) during total knee arthroplasty, is the first to address
this problem specifically and the first to demonstrate the results
of primary repair without use of a more constrained component design.
In a previous study of total knee arthroplasty in fifty morbidly
obese patients, 8% (four) of the patients sustained intraoperative
iatrogenic injury of the medial collateral ligament5; the authors of that study reported
that at their institution no patient with a normal body-mass index
sustained this complication during surgery. Those investigators
opted to reattach the ligament with use of a staple in three patients,
and they used an unlinked constrained implant in the fourth patient;
late instability did not develop in any of the four patients. That
report did not state whether any of the patients had preoperative
valgus deformity that might have predisposed them to incompetence of
the medial collateral ligament. In our series, the mean body-mass
index was 32.5 (obese); however, more than half (eight) of the fourteen
patients had a body-mass index in the normal range. We limited our
study to patients with a neutral or varus preoperative mechanical
axis, in order to eliminate the possibility that the deformity itself,
rather than an actual intraoperative event, resulted in incompetence
of the medial collateral ligament. While we agree that obesity increases
the difficulty of obtaining surgical exposure5,
our experience suggests that this surgical complication may also
occur in patients with a normal body habitus.
Few investigators have commented on the appropriate treatment
for intraoperative injury of the medial collateral ligament; reports
that have addressed this problem have recommended using a more constrained
implant1,3,13,20. Some investigators
have described advancements or reconstructions of the medial collateral
ligament, which may be used with minimally constrained or unconstrained
implants6,7,9. However, these
procedures all have been performed in patients with preoperative
valgus deformity and preexisting weakness, attenuation, or incompetence
of the medial collateral ligament. No reports, to our knowledge,
have advocated direct primary repair or reattachment for the treatment
of intraoperative disruption of the medial collateral ligament in
a varus knee.
Unlinked constrained prostheses transmit increased stresses to
the bone-cement and implant-cement interfaces2,11,13.
This is expected to result in higher rates of aseptic loosening
than are found with less-constrained implant designs1,22,23. Implanting unlinked constrained
prostheses in the primary setting also requires removal of considerably
more femoral intercondylar bone. In addition, constrained implants
are designed for use with a stem. Both of these design features
decrease the remaining bone stock available for revision, should one
be required in the future. The tibial intercondylar eminence of
semiconstrained devices can be an additional source of polyethylene
wear debris and delamination14.
Failure or fracture of the tibial intercondylar eminence24 and recurrent instability despite
an intact intercondylar eminence25 both
have been described with constrained total knee designs. In one
series, valgus instability occurred despite an intact tibial post
in four of five patients with deficiency of the medial collateral ligament
managed with an unlinked constrained device26.
For these reasons, most arthroplasty surgeons are reluctant to use
unlinked constrained prostheses when a more conservative option
is available1,3,4,13,21,22,25.
Since medial-side soft-tissue reconstructions have shown some
promise in arthroplasty of valgus knees6-10,
and since a number of clinical studies in the non-arthroplasty literature
have suggested that the medial collateral ligament has considerable
capacity to heal following injury15-18,27,
we have been performing primary repair or reattachment of the medial
collateral ligament in the uncommon cases in which the ligament
is inadvertently injured during total knee arthroplasty. In the present
study, we evaluated the results of this approach, identifying all
cases of injury of the medial collateral ligament from a prospectively maintained
database of all primary total knee arthroplasties performed at one
institution during the study period.
Evaluation at a minimum of two years after the operation, with
no loss to follow-up, revealed successful results after primary
repair or reattachment of the medial collateral ligament, without
the use of a constrained total knee prosthesis, combined with six
weeks of postoperative bracing. No patient required bracing beyond
the initial six-week period, no patient had subjective or objective
coronal-plane instability, there were no revisions for any reason, and
at a mean of nearly four years the mean Hospital for Special Surgery
knee score was 93 points (excellent).
Nevertheless, intraoperative injury of the medial collateral
ligament is a potentially severe complication. Repair or reattachment
of the ligament obviously increases surgical time. One recent report suggested
that an unstable failed total knee prosthesis is quite resistant
to treatment26. In the present
series, one patient required postoperative manipulation under anesthesia
and had a persistent flexion contracture; this may have been the
result of difficulty with the performance of range-of-motion exercises
while the hinged knee brace was worn.
This series excluded patients with preoperative valgus deformity,
as such patients may have preexisting injury or attenuation of the
medial collateral ligament6,9.
We do not believe that the results of the present study apply to
total knee arthroplasty in valgus knees with attenuation or injury
of the medial collateral ligament. Other investigators have described a
variety of medial-side reconstructions for such patients6-9; those reconstructions should be
considered, perhaps along with use of a more constrained implant.
One limitation of the present study is the absence of objective
techniques to measure joint stability. To our knowledge, there are
no established normal ranges of coronal-plane stability as measured
with a mechanical arthrometer in a knee with a prosthetic replacement.
On the basis of the physical examination with the knee in full extension
and in 30° of flexion, the patients in this series appeared similar
to other patients managed with primary total knee arthroplasty,
without injury of the medial collateral ligament.
We favor using a posterior-cruciate-sparing implant in routine
primary total knee arthroplasty. The high proportion of posterior-cruciate-sparing
implants in this series may have favorably affected the results, as
the posterior cruciate ligament functions as a secondary stabilizer
(after the medial collateral ligament) to valgus stress in the coronal
plane28,29. Twelve of the sixteen
knees in this series were treated with a posterior-cruciate-sparing
prosthesis. These implants had a flat-on-flat articular surface geometry,
so it is unlikely that any coronal-plane stability was provided
by the implant itself.
The mean duration of follow-up in this series was forty-five
months (range, twenty-four to ninety-five months), and there was
no apparent deterioration of results with time. Accordingly, we
remain cautiously optimistic about this method of treatment. Optimally,
injury to the medial collateral ligament should be prevented during
surgery. Direct visualization of the ligament during medial subperiosteal elevation
of the soft tissues, sound knowledge about the location of the origin
and insertion of the medial collateral ligament, and careful protection of
the ligament during tibial and femoral preparation all can decrease
the likelihood of intraoperative injury. Recognition and treatment
of intraoperative injury of the medial collateral ligament, when
it does occur, is of paramount importance in order to avoid the
subsequent need for a revision because of coronal-plane instability.
This study suggests that, as long as a preexisting valgus deformity
is not present, primary repair or reattachment of the medial collateral
ligament can restore stability and allow the use of an unconstrained
total knee prosthesis.