Loosening is one of the causes of failure of a total elbow
arthroplasty1,2.
Often, the loosening is associated with bone destruction, osteolysis, and
cortical thinning or ballooning of the cortex. The options for reconstruction
following loosening of a total elbow replacement include reimplantation of a
new prosthesis without bone reconstruction and with or without use of custom
implants3-5,
implantation of a new prosthesis with impaction allograft
bone-grafting6,
implantation of a new prosthesis with allograft cortical
strut-grafting7, and
the use of an allograft-prosthesis
composite8,9.
Little has been published about any of these options, and virtually nothing
has been written about composite allograft-implant reconstruction, to our
knowledge. The purpose of this study was to present our experience with this
technique in thirteen revision procedures.
Patients
Between 1990 and 2000, thirteen patients (thirteen elbows) with a failed
total elbow arthroplasty were operated on with use of an allograft-prosthesis
composite. Four of the procedures were performed with a humeral
allograft-prosthesis composite and nine, with an ulnar allograft-prosthesis
composite. A Coonrad-Morrey semiconstrained total elbow arthroplasty was used
in all but two patients, in whom a Pritchard-Walker Mark-II linked implant was
inserted. Of the original cohort of thirteen patients, two died from unrelated
causes, twenty-four months and seventy-two months postoperatively. The results
at their last follow-up evaluation were included in the study. The thirteen
patients were followed for an average of forty-two months (range, twenty-four
to seventy-two months). There were four men and nine women. The mean age at
surgery was sixty-two years (range, thirty-nine to seventy-seven years). The
right elbow was operated on in five patients and the left elbow, in eight. The
dominant arm was operated on in five patients.
The total elbow arthroplasty was initially indicated for rheumatoid
arthritis in six patients and for posttraumatic arthritis in seven. An average
of 1.7 prior arthroplasties (range, one to four) had been performed on the
thirteen patients before the allograft-prosthesis composite procedure: seven
patients had had one such procedure, four had had two, one had had three, and
one had had four. The prosthesis that was revised with the
allograft-prosthesis composite procedure was a Pritchard-Walker implant in
four patients, a GSB-I (Gschwend-Scheier-Baehler-I) implant in one, a
Coonrad-I implant in one, a Coonrad-Morrey implant in four, and a custom
implant in three. The interval between the last total elbow arthroplasty and
the allograft-prosthesis-composite procedure averaged eight years (range, one
to twenty-three years). The data characterizing the thirteen patients are
summarized in Table I.
Assessment
All patients were followed with use of the Mayo Clinic Total Joint
Database. Preoperative data were obtained from the records, and the patients
were contacted postoperatively and asked to complete a questionnaire and to
return for a clinical and radiographic evaluation. Radiographic evaluation was
performed for all patients, at a mean of forty-two months (range, twenty-four
to seventy-two months).
Eight of the thirteen patients were examined at the Mayo Clinic, and five
were evaluated by their local orthopaedic surgeon. The preoperative and
postoperative Mayo Clinic elbow assessment form was completed by an
orthopaedic resident or a physician-assistant (R.A.A.), and the questionnaire
was completed by the patient. A handheld goniometer was used to measure the
arcs of elbow flexion and forearm rotation. All of the radiographs were
reviewed in a nonblinded fashion by two of us (P.M. and R.A.A.), and a
consensus was reached. Final standardized anteroposterior and lateral
radiographs were made for all patients.
A final evaluation of subjective and objective data was performed to allow
the calculation of the Mayo Elbow Performance Score (MEPS), which assigns a
maximum score of 45 points for pain, 20 points for motion, 10 points for
stability, and 25 points for daily functional
activities10. On
the basis of this score, the results were defined as excellent (90 to 100
points), good (75 to 89 points), fair (60 to 74 points), or poor (<60
points). In addition, adequate preoperative data were available to allow
calculation of a preoperative MEPS for all thirteen elbows, and this was done
at the latest follow-up evaluation for all of the elbows. On radiographic
evaluation, four features were particularly documented: the union of the
allograft to the host bone, the presence or absence of resorption of the
distal aspect of the humerus or the proximal aspect of the ulna, and the
presence and extent of any radiolucency. Wear of the bushings on the
anteroposterior radiograph was assessed according to the method described by
Gill and
Morrey10.
Indications
The indication for the allograft-prosthesis-composite procedure was aseptic
loosening with bone deficiency. Elbows with a known infection or with a
positive culture of intraoperative specimens were excluded. Preoperative bone
loss in the distal aspect of the humerus was classified as Type I when the
defect involved the bone around the articular part of the previous total elbow
replacement up to the olecranon fossa, Type II when it involved the distal
third of the humerus around the stem of the previous prosthesis, and Type III
when it involved the humerus proximal to the stem of the previous prosthesis.
Preoperative bone loss in the proximal aspect of the ulna was classified as
Type I when the defect involved the olecranon process with the triceps tendon
attachment, Type II when it involved the proximal third of the ulna around the
previous prosthesis, and Type III when it involved the ulna distal to the
previous prosthesis. A humeral defect was present in seven elbows, and the
defects averaged 8 cm (range, 4 to 15 cm) in length. A defect of the ulna was
present in twelve elbows, and the average length of those defects was 4.5 cm
(range, 2 to 10 cm). A humeral allograft-prosthesis composite was selected for
four elbows, with a distal humeral bone defect averaging 10.5 cm, and an ulnar
allograft-prosthesis composite was chosen to reconstruct a bone defect of the
ulna in nine elbows (Table
II).
Technical Considerations
Surgical data are summarized in Table
III. The operations were performed with the patient supine with a
sandbag under the scapula. In eight cases, the ipsilateral iliac crest was
prepared and draped. The previous posterior skin incision was always used.
After the exposure of the medial aspect of the triceps, the ulnar nerve was
always identified with use of ocular magnification and a bipolar cautery. It
was dissected free from scar tissue. In eight patients, the triceps was then
released from the olecranon and reflected laterally in continuity with the
anconeus according to the method described by Bryan and
Morrey11. In four
patients, the triceps was split in two parts according to the method described
by Gschwend et
al.12. In one
patient, the triceps was left intact on the olecranon, and the distal part of
the humerus was exposed through both the lateral and the medial border of the
triceps.
The radial nerve was isolated only when the defect was mainly on the
humeral side, in order to proceed with placement of a humeral allograft or
with cement removal with the radial nerve under control. The nerve was
isolated in seven patients.
The joint was then exposed and disarticulated. Both components were
inspected. Loose implants were removed, but well-fixed components were left in
place. Four humeral components (two Coonrad-Morrey and two Pritchard-Walker
implants) were left in place, and all ulnar components were removed. The
cement was then removed with use of an ultrasonic device (Ultra-drive; Biomet,
Warsaw, Indiana). All synovitic membrane that contained metal debris was
carefully débrided. Pathologic specimens were always sent to the
laboratory for histologic assessment and for culture. No sign of acute or
chronic infection was found at the time of any procedure.
The osseous defect was then evaluated. When a patient had a humeral defect
up to 5 cm proximal to the olecranon fossa and a noncustomized implant could
not be used, an allograft-prosthesis composite was chosen
(Figs. 1-A, 1-B, 1-C, 1-D). The
medullary canal of the proximal fragment was identified with use of a
high-speed burr and reamer, and step-cuts were performed in the humerus to fit
the allograft. The medullary canal of the allograft was identified, and a
cutting-guide was used to fit the humeral component perfectly. A 15-cm-long
humeral implant was used in two patients, and a 20-cm-long humeral implant was
used in two. The humeral prosthesis was fitted to the allograft and, after a
good fit to the host was confirmed, it was cemented into the allograft with
vancomycin-impregnated cement. The prosthesis was then cemented into the host
and stabilized by additional fixation. Radiographs were used to ensure that
the humeral component was in the medullary canal and had not violated the
cortex. An iliac crest corticocancellous bone graft was placed around the
junction between the host and donor bone of the humerus in two patients, and
it was secured with number-5 nonabsorbable (Mersilene) sutures. Additional
cortical allograft was placed at the anterior aspect of the humerus and
beneath the anterior flange of the humeral component.
When a patient had a defect of the ulna involving all of the olecranon as
well as the proximal third around the previous implant, the ulnar bone did not
provide adequate fixation distally, and reconstruction of the ulnar bone was
not possible proximally, an ulnar allograft-prosthesis composite was chosen
(Figs. 2-A, 2-B, 2-C, 2-D,
2-E). The subcutaneous border of the ulna was exposed, and the
incision was extended distally to expose the normal part of the ulna. The
debris from the ulnar canal was removed with great care, and the medullary
canal was identified. Serial reaming and rasping was performed in order to
allow introduction of the ulnar component. A long-stem ulnar component (114
mm) was used for eight patients. No custom devices were employed. The
allograft was step-cut in such a way as to place the ulnar component into it
and the remaining portion of the ulnar component into the medullary canal of
the host bone. After the ulnar medullary canal had been prepared the ulnar
component was cemented into the allograft, and after the cement had cured it
was inserted into the host medullary canal. Cancellous bone graft was packed
around the allograft. The step-cut usually provided excellent rotational
stability; nonetheless, a seven or eight-hole semitubular side-plate was
applied in four patients, and two or three cerclage wires were used for the
other five patients.
Trial reduction was essential to evaluate the arc of motion and to identify
any impingement with the allograft. Sometimes, a portion of the medial and/or
lateral condyles had to be removed in order to achieve a smooth articulation.
On the ulnar side, a prominent coronoid or olecranon was usually excised.
After careful hemostasis, the triceps was secured, usually at 90° of
flexion, with two number-5 nonabsorbable Mersilene sutures placed through
drill holes in the proximal part of the ulna. In the patients with an ulnar
allograft-prosthesis composite, drill holes were placed across the proximal
part of the ulna to reattach the
triceps11. Residual
autogenous bone, with its attached soft tissue, was wrapped around the
allograft-host bone junction as a vascularized graft. The operative time
averaged 4.7 hours (range, 3.5 to six hours).
The elbow was then held in 90° of flexion with an anterior plaster
slab. The dressing was changed at two to three days. Motion of the elbow was
started at two or three days for twelve patients but was delayed for two. It
was delayed for one patient because a Pritchard-Walker implant had been used
and for the other because of a weak triceps. Strengthening exercises were
avoided. Patients were advised not to lift >0.5 to 1 kg during the next
three months and were told to avoid lifting >1 kg on a repetitive basis or
5 kg as a single event permanently.
Overall Outcome
At an average of forty-two months (range, twenty-four to seventy-two
months) postoperatively, the MEPS was excellent for four elbows, good for
three, fair for one, and poor for five. The average preoperative score,
calculated retrospectively, was 23 points (range, 5 to 45 points), and the
average postoperative score was 67 points (range, 15 to 100 points). The data
characterizing each patient are summarized in
Table IV. Seven of the thirteen
patients were satisfied with the result of the surgery.
Pain
Preoperatively, ten of the thirteen patients had severe pain in the elbow;
two, moderate pain; and one, mild pain. At the most recent follow-up
evaluation, five patients had no pain in the elbow; four, mild pain; and four,
moderate or severe pain. The average score for the pain component of the MEPS
improved from 5 points preoperatively to 30 points at the most recent
follow-up evaluation.
Range of Motion
The average preoperative arc of flexion was 87°. The average
postoperative arc of flexion was 97°, with an average of 28° (range,
0° to 60°) of extension and 125° (range, 100° to 140°) of
flexion. Five patients achieved a functional arc of motion. The average
preoperative arc of forearm rotation was 140°. Postoperatively, the
average arc of forearm rotation increased slightly, with an average of 72°
(range, 50° to 85°) of pronation and an average of 72° (range,
50° to 80°) of supination.
Stability
Eight of the thirteen elbows were stable at the time of final follow-up.
Two elbows became moderately unstable because of a deep infection and nonunion
of the allograft, and three were grossly unstable because the
allograft-prosthesis composite had to be removed to treat a deep infection
(two elbows) and because of nonunion (one elbow).
Daily Function
Preoperatively, the ability to perform activities of daily living was
severely limited by the unstable elbow. The average preoperative score for the
functional component of the MEPS was 2 points (range, 0 to 10 points).
Postoperatively, the average score was 15 points (range, 0 to 25 points). At
the latest follow-up evaluation, five of the thirteen elbows caused no
difficulties with daily activity. Formal strength studies were not done in
this series.
Radiographic Findings
Nine elbows showed complete incorporation of the allograft to the host
bone. Two had evidence of nonunion of the allograft, both on the humeral side.
The allograft-prosthesis composite had to be removed because of a deep
infection from the remaining two elbows, two months and three months after the
initial procedure, and the allografts were not yet incorporated at the time of
writing.
Of the nine elbows with incorporation of the graft, none had progressive
radiolucency around either the humeral or the ulnar component. The bone graft
behind the anterior flange of the Coonrad-Morrey implants had matured with
complete incorporation (Fig.
2-E). None showed radiographic evidence of bushing wear. No sign
of resorption of the allograft was observed at the latest follow-up
evaluation. In one patient, a violation of the cortex at the tip of an ulnar
component was seen with no sign of loosening
(Fig. 1-C).
Complications and Revisions
There were six complications affecting six elbows, five of which were
revised.
Deep infection developed in four elbows, one, two, three, and twenty-four
months after the allograft-prosthesis-composite procedure. In the first
patient, a wound dehiscence appeared three months postoperatively. The wound
was débrided and irrigated with antibiotic solution before the humeral
allograft-prosthesis composite was removed because of persistence of the
infection. Cement with antibiotics was left in the wound, and two additional
débridements were performed. At the time of follow-up at thirty-six
months, the patient had no pain and had normal motion. The elbow was stable
but weak. Radiographs showed consolidation of the allograft at the host-bone
junction. There was no humeral lucency, but the ulnar component had fractured
and had been replaced. One year later, the elbow was moderately unstable with
moderate pain and weakness limiting daily activities.
In the second patient, an acute infection developed two months
postoperatively. The allograft-prosthesis composite and all of the cement was
removed, and the elbow was placed in a brace. The patient died of unrelated
causes two years later.
A deep infection developed one month postoperatively in the third patient.
A surgical débridement was performed, the cerclage wires and joint
bushings were removed, and the joint was disarticulated. After another
débridement, the findings on culture were negative. The total elbow
replacement was rearticulated one month later. One year later, the result was
satisfactory with no recurrence of infection.
Finally, an infection developed two years after insertion of an ulnar
allograft-prosthesis composite in the fourth patient, who had rheumatoid
arthritis and in whom multiple joint infections had developed from an ingrown
toenail. The elbow was débrided. The humeral component was stable, but
the ulnar component was loose in the allograft and had to be removed. The
infection resolved, and the elbow was protected in a hinged brace at the time
of writing.
One patient sustained a humeral fracture associated with a loose humeral
component and nonunion of the allograft-prosthesis composite, three years
postoperatively. The humerus was reconstructed with a tibial strut and
impaction grafting. A 20-cm humeral component was inserted and articulated
with the previous ulnar component. Three months later, a deep infection
developed and the total elbow replacement and grafts were removed.
Another patient had an obvious nonunion at the allograft-humerus junction
on radiographic examination. The result was fair, but because the patient had
only slight pain and a functional arc of motion and no implant loosening on
radiographic examination, no revision procedure was performed.
Two patients had ulnar nerve symptoms prior to surgery, and these were
essentially unchanged after the procedures.
Revision of a failed total elbow arthroplasty can be a challenging
procedure. Treatment options usually depend on the age of the patient, the
diagnosis, the anticipated use of the extremity, and the characteristics of
the failure mode. Loosening of the implant appears to be the most frequent
cause of failure of total elbow
arthroplasty1-3,13,14.
If the failure requires removal of the components, and a reimplantation option
has been chosen, resorption of a substantial amount of bone can pose major
problems during any reconstructive procedure. Since bone stock is rarely
preserved, resurfacing implants usually cannot be
used15,16.
When there is bone deficiency or a grossly unstable elbow, a semiconstrained
implant becomes the treatment of
choice3,17.
The Coonrad-Morrey total elbow replacement seems to be a satisfactory
option for addressing the resected joint after failure of a total elbow
arthroplasty or after
trauma18-22.
The range of sizes and lengths of the humeral and ulnar stems and the flange
can address many situations, even a case with up to 8 cm of distal humeral
bone deficiency5.
When bone loss is greater, a custom implant, a component reconstruction with a
strut graft, or an allograft-prosthesis composite is each an option. An
allograft or an allograft-prosthesis composite is indicated for massive bone
loss8,9.
In our series, an allograft-prosthesis composite was chosen for the humeral
side because of bone defects averaging 10.5 cm in length, and it was chosen
for the ulnar side because of bone defects averaging 4.5 cm in length. In both
cases, the defects exceeded the length of the standard stem of the
component.
Cadaveric elbow allografts have been used to restore destroyed elbow joints
since the beginning of the century. Lexer reported the first clinical
experience in
192523, but it was
not until the 1970s that this option was used routinely in the treatment of
bone tumors24. The
results were encouraging, and massive allograft-prosthesis composites were
then used to salvage failed hip and knee
replacements9,25-31.
In 1987, Dee and Ries mentioned allografts as an option for the treatment
of failed elbow replacement but provided no
details32. Other
authors have pointed out that a structural allograft may be used for extensive
traumatic defects of the
elbow33,34
or for a salvage procedure after failed total elbow
arthroplasty35,36.
In 1987, Urbaniak and
Aitken37 reported
on twelve patients, two of whom had been treated with a distal humeral
allograft and ten of whom had been treated with a total elbow
prosthesis-allograft combination for posttraumatic arthritis of the elbow.
Nonunion occurred in four patients because of inadequate stabilization of the
allograft to the host bone. Deep infection requiring removal of the allograft
developed in two patients because of poor skin about the olecranon. Finally,
radiographic evaluation showed that most of the allografts had undergone
degeneration with time. Complications occurred much more frequently in
association with the humeral allografts than they did in association with the
ulnar allografts. More recently, Dean et
al.38 performed
elbow allograft reconstruction to treat massive posttraumatic bone loss in
twenty-three patients. Ten of fourteen patients who had been followed for an
average of 7.5 years reported a satisfactory result. However, complications
occurred in sixteen of the twenty-three patients, and the allograft had to be
removed in six (because of infection in two, instability in three, and
nonunion in one).
There are few reports in the literature concerning the results of the use
of an allograft-prosthesis composite for any joint. In a review of the results
of 718 procedures involving the use of a massive allograft after tumor
resection, 109 of which involved use of an allograft-prosthesis composite,
Mankin et al.8 found
that twelve procedures were followed by infection; twenty-one, by a fracture
of the allograft; nineteen, by a nonunion between the allograft and the host
bone; and six, by an unstable joint. Of the 109 procedures involving an
allograft-prosthesis composite, seventy-five had an excellent or good result.
To our knowledge, Gross and
Hutchison30
reported the most extensive experience with allograft-prosthesis composites
for nontumorous conditions. They reviewed the results of 200 procedures in
which a circumferential allograft was used for revision of a failed total hip
arthroplasty with massive bone loss. They reported that 170 (85%) of the
operations were successful at a mean of 4.8 years. There were twenty-five
reoperations (12.5%): eleven were performed because of dislocation; six,
because of infection; seven because of nonunion; and one, because of
loosening. No graft was revised because of resorption. A review of the
literature concerning the use of allograft-prosthesis composites for revision
of failed total hip and total knee arthroplasties revealed rates of
satisfactory results ranging from 55% to 85% and revision rates ranging from
10% to
45%25-31.
In our small series of procedures at the elbow, the rate of satisfactory
results was 54% and the revision rate was 38%. The allograft-prosthesis
composite seems an inferior construct compared with its effectiveness at other
joints25-31
but better than allograft
alone37,38.
Infection developed in four of our patients. Three of the four infections
developed in elbows with posttraumatic arthritis and previous surgical
procedures. Healthy pliable skin and subcutaneous tissue can decrease the risk
of infection8.
Furthermore, we now routinely incorporate 1 g of vancomycin in every 40 g of
cement at the time of reinsertion of the implant.
Nonunion between the host and the allograft is also a frequent
complication, developing in
3%26,30
to 22%9 of cases.
Previous studies have emphasized the necessity of obtaining good initial
stability of the prosthesis in the allograft and the host bone to avoid this
complication. Creating a step-cut junction seems important for achieving this
stability on the humeral side. Also important is the use of a long-stem
prosthesis that bypasses the allograft-host bone junction. Gross et
al.26,30
as well as Clarke et
al.29 cemented the
prosthesis only into the allograft and obtained press-fit fixation of the
prosthesis into the host bone. Cement should be kept away from the
allograft-host bone
interfaces28. Two
nonunions were observed in our series. One was associated with a loose humeral
implant and a periprosthetic humeral fracture and required a revision. The
other nonunion was diagnosed on radiographic examination.
No nonunions were seen in the patients treated with an ulnar allograft,
perhaps because stability of the allograft-prosthesis composite was obtained
more readily on the ulnar side than on the humeral side, with press-fit
fixation of the component into the allograft and the host bone being more
common on the ulnar side than on the humeral side. Moreover, rigid fixation
with a plate or cerclage wires was performed for all of the ulnar
allograft-prosthesis composites whereas supplemental fixation was performed
for only two of the four humeral allograft-prosthesis composites.
A fracture of the allograft occurred in two of the eleven patients in the
study by Clarke et
al.29. The
fractures were related to failure to bypass the allograft-host bone junction
with a long-stem prosthesis and inadequate use of plate-and-screw fixation.
Both of those factors could create stress risers in the allograft bone. In our
series, this complication occurred in one patient and was associated with a
loose humeral implant and nonunion of the allograft.
In conclusion, an allograft-prosthesis composite can be a valuable option
for selected patients with a failed total elbow arthroplasty associated with
massive bone loss. The union and implant survival rates are high when the
allograft is fixed to the host bone with contemporary techniques. However, the
complication rate is also high, and deep infection can be a devastating
problem that generally requires removal of the allograft, leading to a poor
functional result. Today, the technique is reserved for cases not amenable to
strut-graft reconstruction.