We retrospectively reviewed the records on 3100 patients with hemophilia
who had presented to our institution and on 1358 total elbow replacements
performed at our institution between 1979 and 2001. Five patients had had
total elbow replacement for the treatment of hemophilic arthropathy
(Table I), and we further
analyzed those cases.
The mean age of the five patients was thirty-nine years (range, twenty-five
to fifty-eight years). The elbow on the dominant side was involved in one
patient, the elbow on the nondominant side was involved in one patient, and
both elbows were involved in three patients. The primary hematological
abnormality was hemophilia A (factor-VIII deficiency) in two patients and
hemophilia A and B (factor-VIII and IX deficiency), hemophilia C (von
Willebrand disease), and hemophilia A with factor-VIII inhibitor in one
patient each. All five patients had acquired HIV disease during the course of
their medical treatment for hemophilia, two patients had hepatitis-C
infection, and one patient required regular peritoneal dialysis because of
chronic renal failure. Polyarthropathy was present in all five patients, with
three having involvement of the knee requiring arthroplasty (Cases 2, 3, and
5) and one each having involvement of the hip (Case 2) and shoulder (Case 1).
One patient (Case 4) had bilateral elbow involvement without involvement of
any other joint. The three patients with bilateral elbow involvement had a
unilateral elbow replacement, with the decision to replace one side in
preference to the other based on patient preference, clinical assessment, and
radiographic signs of joint destruction. Preoperative radiographs were
classified with the system of Arnold and
Hilgartner1: three
patients had stage-4 disease and two, stage-5 disease
(Fig. 1-A).
Preoperative preparation involved screening for clotting factor and
antibody. Preoperative factor replacement was prescribed when clotting factor
titers were <100%. We began with a bolus of recombinant factor VIII. With
the units calculated as (100% — patient's baseline factor activity
level) × weight in kg × 0.5, we gave a continuous infusion of 2 to
4 units/kg/hr depending on the severity of disease and the amount of bleeding
encountered at surgery. We then measured factor-VIII levels postoperatively
and daily thereafter and adjusted the infusion dose to keep the serum level at
approximately 75% to 100% of normal. Once the patient was able to walk and/or
ready to leave the hospital, we converted the continuous infusion to daily
bolus dosing.
Patients with inhibitors to factor VIII are treated with the following
protocol. If the inhibitor titer is <5 BU, recombinant factor VII is
administered. If the inhibitor titer is >5 BU, a titer is checked against
porcine factor VIII. If the titer is less than standard, the porcine factor
VIII is used for the infusion. An alternative treatment is to use factor-VIII
inhibitor bypassing activity (FEIBA) at 72 units/kg given twice a day on the
day of surgery and as needed, tapering to once a day. A final option is to
administer recombinant factor VIIa, 90 µg/kg, every two hours until
hemostasis is achieved and then every three to six hours for several days
after surgery.
Because of the risk of infection to the operating staff, the preoperative
setup involved the surgeon and the first assistant setting up two Mayo stands
for their respective sides of the operating table with instruments to be used
during the procedure. It was planned that very little instrument passing would
occur among staff. Other standard precautions for dealing with the high risk
of infection to the operating staff were used; these included wearing face
shields, using double gloves, placing used sharp instruments on a tray, and
avoiding hand-to-hand transmission.
Operative Technique
A standard posterior approach, under tourniquet control, was used in all
patients. The ulnar nerve was identified in all patients but one, in whom the
triceps was reflected from lateral to medial. Elbow contractures were released
by capsulectomy and/or capsulotomy, a complete synovectomy was performed in
three patents, and the bones were prepared in the standard manner. Case 3 was
exceptional in that both the humerus and the ulna had an obliterated,
sclerotic medullary canal, which required the use of high-speed drills (Midas
Rex, Fort Worth, Texas) to refashion a canal capable of receiving the
arthroplasty stems. Another patient (Case 1) had very small medullary canals
that required careful reaming to avoid breaching the cortex. When bone and
canal preparations were complete, the tourniquet was released and stringent
hemostasis was secured. The tourniquet was then reinflated, the canals were
dried, cement was injected in a semiviscous state, and the implants were
inserted in the standard manner. The Coonrad-Morrey total elbow replacement
(Zimmer, Warsaw, Indiana) was used in four patients
(Fig. 1-B), and a Mayo modified
Coonrad implant (Zimmer) was used in one (Case 3). The length of the humeral
stem was chosen on the basis of the amount of bone stock that had been
preserved. With a normal bone length and an adequate canal, a 4-in (10.2-cm)
stemmed implant was inserted. If the bone length was deficient and/or the
canal was capacious, a 6-in (15.2-cm) stem was chosen.
The average operative time was one hour and forty-six minutes (range, one
hour and thirty-nine minutes to one hour and fifty-five minutes). The elbow
was splinted in extension for the first postoperative day, and the patient
performed passive motion thereafter.
Follow-up
The patients were seen clinically at a mean of 5.8 years (range, one to
eleven years) after surgery, and the Mayo Elbow Performance Score (MEPS) was
calculated from the data obtained at that evaluation. Information obtained
with a telephone and/or mail questionnaire allowed additional follow-up, to an
average of 10.2 years (range, seven to thirteen years) after surgery. Two
patients died (at seven and thirteen years, respectively) after the index
total elbow replacement; one (Case 1) died of AIDS and the other (Case 3), of
chronic renal failure. Both had a functional joint at the time of death
according to their families.
Functional Scores (Table
II)
The Mayo Elbow Performance Score (MEPS) was used to assess the functional
changes resulting from the surgical intervention. One patient (Case 4) was not
included in this analysis because the prosthesis had been removed, as a
consequence of deep infection, two months postoperatively. The four remaining
patients, in whom the implants remained in situ, had a marked decrease in
pain, with an improvement in the average pain score from 4 (range, 0 to 15) of
45 points preoperatively to 41 points (range, 30 to 45 points) at the final
evaluation. The average score for stability also improved, from 5 of 10 points
to 10 of 10 points. The average flexion arc increased from 63° to 101°
preoperatively to 25° to 140° intraoperatively to 40° to 129°
at the final evaluation. The overall MEPS increased from 24 of 100 points
preoperatively to 90 of 100 points at the final evaluation.
Radiographic Analysis
Final follow-up radiographs revealed the positions of the components to be
adequate in the four patients in whom the implants were in situ. Two patients
had no radiolucent lines at either the bone-cement or the cement-implant
interfaces (Fig. 2), one (Case
1) had a radiolucent line over <25% of the ulnar cement-implant interface,
and one (Case 3) had a radiolucent line over 75% of the ulnar bone-cement
interface and over 50% of the humeral cement-implant interface
(Fig. 2).
Complications
Three of the five patients had a complication, with two requiring
additional major surgical intervention. One (Case 4) was returned to the
operating room during the evening of the index total elbow arthroplasty
because of uncontrollable hemorrhage from the operative site. At exploration,
no large-vessel source of the bleeding was found. The wound was packed and
compressed, hemostasis was achieved, and the wound was closed. Two months
postoperatively, the posterior skin became necrotic and a deep wound infection
was diagnosed. The implant could not be salvaged despite multiple wound
débridements, and it was removed. The wound infection, which was caused
by Bacillus species, was controlled with antibiotic packs and repeated
débridement, and the posterior wound was eventually closed with a
pedicled flexor carpi ulnaris muscleflap split-thickness skin graft. This
patient retained good arm function with the permanent use of a hinged brace;
the flexion arc was 30° to 105°.
A deep infection developed four years after the surgery in another patient
(Case 1), presumably as a result of a hematogenous mechanism. The wound was
opened, the implant was disarticulated, the wound was débrided and
packed with an antibiotic spacer, and additional systemic antibiotics were
given for two months, since the organism isolated was Staphylococcus
aureus. When the systemic inflammatory markers returned to normal, the
wound was found to be free of infection, and the implant, which was intact,
was rearticulated. The elbow remained problem-free until the patient's death
due to end-stage HIV disease three years later.
One patient (Case 5) had persistent but poorly localized pain in the elbow
following the arthroplasty. Recurrent hemorrhage into the elbow was noted in
one patient (Case 3), and although it was not considered to be a complication
attributable to the surgery, it is mentioned here for completeness. Neither of
these two patients needed additional surgical intervention.
The elbow is the predominant site of arthropathic involvement in the upper
limb of patients with
hemophilia1,6,7.
The elbow is involved in
25%8 to
87%7 of cases, but
rarely is it the only joint
affected9. The
relative frequency of involvement of the dominant and nondominant elbow has
previously been
debated9,10.
The coexistent relationship between knee and elbow pathology noted in the
literature9 appears
to be a loadsharing issue. Repetitive increased load-bearing by the elbow is
thought to be responsible for recurrent hemarthroses and subsequent
arthropathy. Interestingly, one of the patients in the present report had sole
involvement of both elbows that cannot be wholly explained by this
hypothesis.
When conservative and medical management fails and symptoms persist,
various surgical options can be considered. Synovectomy is the most frequently
performed surgical
intervention11-14.
The prevalence of recurrent hemarthrosis is reduced after synovectomy, and
there is some evidence that progression of arthropathy can be delayed by early
synovectomy13,15.
Our experience appears to agree with these observations, as only one patient
(Case 3) of five had recurrent intra-articular hemorrhages following the total
elbow arthroplasty and its obligatory synovectomy. Interposition
arthroplasties performed for severe hemophilic arthropathy at a single center
were reported to have moderate success after an average duration of follow-up
of six years16.
Hemiarthroplasty has also been described, but it resulted in elbow ankylosis
in both patients in whom it was
performed17.
Total elbow replacement in patients with rheumatoid arthritis may achieve
medium and long-term success rates on par with those of hip
replacements18,19.
However, total elbow replacement for hemophilic arthropathy is not a common
procedure and has not been well documented in the literature. Our review of
the literature reminded us of the relatively recent evolution of elbow
replacement. In the majority of the published case reports, the patients were
treated with a rigid hinged
design5,20,21,
but those cases can be largely overlooked since we are now aware of the design
flaw, with respect to loosening, of a rigid hinged elbow
replacement22. All
of the patients in our current study were treated with a linked,
semiconstrained total elbow replacement. While the published literature
concerning total elbow replacement in patients with hemophilia is generally
not encouraging because of the high rates of
complications20,23,
our current study is a reason for greater optimism, as the functional results
were good (one patient) or excellent (three patients) at an average of ten
years.
This form of surgical intervention should still not be considered routine
or straightforward. The surgical work-up and technique are of paramount
importance, in order to anticipate problems and to reduce the risk of disease
transmission to the operating staff, since HIV disease is a common feature of
hemophilia. The preoperative surgical work-up involves screening for all
clotting factor titers and their inhibitors and obtaining a CD4 count as a
screen for HIV
infection24. If the
clotting factor titers are low, a preoperative transfusion to raise the level
beyond 100% allows surgery to proceed safely. However, the case of one of our
patients, with factor-VIII inhibitor, highlights the cause for concern posed
by this relative contraindication to surgery. The inhibitor prevented regular
attempts to raise factor-VIII titers and was thought to be responsible for
severe immediate postoperative hemorrhage, which required additional surgical
treatment. The consequence of this poor hemostatic control was skin flap
necrosis and deep-seated infection, culminating in implant removal. However,
the patient subsequently had good function following a resection arthroplasty.
This sequence of complications has been previously reported after total knee
replacements in patients with
hemophilia25. In
order to minimize the risk of this complication, some authors have used
recombinant factor VIIa, with good results, when performing major joint
replacements in patients with factor-VIII
inhibitor4,26.
With greater experience, this intervention may eliminate the presence of the
inhibitor as a relative contraindication to joint replacement.
Intraoperative preparations should account for abnormal or difficult
anatomy. We treated a patient with a very small medullary canal requiring an
extra-small ulnar component and appropriately small reamers. Another patient
had obliterated medullary canals, with sclerotic bone, a feature that should
be specifically sought on preoperative radiographs.
The rates of complications after joint replacement in patients with
hemophilia have previously been reported. Ragni et
al.27 concluded
that, in patients with HIV disease, the sites of knee operations become
infected three times more often than do the sites of hip operations in this
population. Rates of deep infection associated with total knee replacements
have been variously reported to be
6%28,
13%29, and
29%27,30.
The rate of deep late infection at the sites of total hip replacements has
been reported to be
9%31. We are not
aware of any available data on infection from a published series of total
elbow replacements. A deep infection developed in two of our five patients;
one infection developed early, requiring implant removal, and the other
developed late and was successfully treated with wound débridement and
antibiotics.
An aseptic loosening rate of 36% was reported at an average of 7.6 years
after hip
replacement32. Two
of our four patients with the elbow implants in situ at the time of follow-up
had some degree of radiolucent lines, although only one of them had a
radiographic appearance that was truly classifiable as aseptic loosening with
progressive lucencies (Fig.
2).
Quality of life is a paramount factor to be considered when deciding
whether to perform a total elbow replacement in a patient with hemophilia with
or without HIV disease. The risks and benefits of this intervention must be
considered by both the surgeon and the patient. The preoperative work-up
should include raising the clotting factor levels to normal, and the presence
of factor-VIII inhibitor should be viewed as a relative contraindication,
after the algorithm for dealing with this issue has been considered. Operative
preparations should include making sure that options to deal with sclerotic,
obliterated, and very small medullary canals are available. The functional
improvements following total elbow replacement with a linked, semiconstrained
prosthesis are reliable, but a high complication rate should be expected.