The knee is the most commonly involved joint in hemophilia and the
one most responsible for long-term disability. In end-stage disease,
pigment-stained hyper-plastic synovial tissue undergoes metaplasia to fibrous
tissue, resulting in contractures that eventually lead to fibrous
ankylosis1-3.
Incapacitating pain and impaired function as a result of extensive erosion of
the joint surfaces and a severely restricted range of motion are the usual
indications for total knee replacement in hemophilic patients. Performance of
a total knee replacement is complex in these patients because of severe
arthrofibrosis, deformity, and bone stock deficiency.
Although pain relief and a decrease in disability have been documented in
the short term in hemophilic patients who have undergone a total knee
replacement4-10,
there is a paucity of data on long-term results. Few studies have included a
mean duration of follow-up of longer than five
years10-16,
and usually the numbers of patients have been small. Although the results of
total knee replacement in these patients have been described as
favorable4,6,7,9,10,15-19,
many authors have reported a high rate of early and late complications, often
related to infection and a restricted range of
motion5,6,14,15,20,21.
It has been hypothesized that the increased risk of infection is due to
impairment of the immune system secondary to infection with the human
immunodeficiency virus (HIV).
The purpose of this study was to evaluate the relatively long-term results
of primary total knee replacement in a large group of hemophilic patients
treated at a single institution by the same primary surgeon.
From April 1975 to September 2001, 103 primary total knee
replacements were performed in seventy-seven hemophilic patients by the same
primary surgeon (J.V.L. Jr.) at a large hemophilia center in the United
States. Of the 103 knees, thirteen were not included in the present analysis:
seven patients (seven knees) had died within two years after the primary total
knee replacement, and six patients (six knees) who lived outside of the area
had been lost to follow-up. The causes of the seven deaths were complications
of AIDS (two patients), an acute cardiac event (one), an intracranial
hemorrhage (one), abdominal bleeding (one), a narcotics overdose (one), and
unknown (one).
Ninety primary total knee replacements in sixty-eight patients who had been
followed for more than two years or until the prosthesis failed (mean duration
of follow-up, 7.8 years; range, 1.4 to 25.7 years) are the focus of this
report (Table I). Of the ninety
total knee replacements, forty-six (51%) were in patients with a contralateral
total knee replacement. After the study had been approved by our institutional
review board, clinical information on these ninety total knee replacements was
retrospectively reviewed. The demographic data that were collected included
age, type and severity of the hemophilia, HIV status and CD4 count at the time
of the surgery, the preoperative and postoperative ranges of motion, and
postoperative complications.
The mean age of the patients at the time of the surgery was 40.1 years
(range, 17.5 to 70.5 years). Sixty patients (88%) had hemophilia A, and eight
patients (12%) had hemophilia B. Eighty-three (92%) of the ninety total knee
replacements were performed in patients with severe hemophilia. Thirty total
knee replacements (33%) were performed in patients who were HIV-negative at
the time of the surgery, and forty-one (46%) were performed in patients who
were HIV-positive at the time of the surgery. Nineteen replacements (21%) were
done in patients for whom the HIV status was unknown at the time of the
surgery, but all of those nineteen patients tested positive in later
evaluations and, for the purpose of this report, were considered to be
positive for HIV at the time of the surgery. The mean preoperative CD4 count
in the forty-one patients who tested positive for HIV at the time of the
surgery was 437.5 cells/mm3 (range, 33 to 1260
cells/mm3).
A thorough preoperative medical and hematological evaluation was performed
for all patients. Clotting factor replacement was infused preoperatively until
a minimum level of 100% of normal was achieved, the level was maintained at
60% of normal for ten to fourteen postoperative days, and then clotting factor
replacement was infused to obtain a level of 30% of normal prior to physical
therapy sessions for six to nine weeks.
The operative procedures were performed in a laminar flow operating room,
under tourniquet control, with standard surgical techniques. Since 1985, all
members of the surgical team have worn self-contained, HEPA (high-efficiency
particulate air)-filtered space suits. Universal precautions as recommended by
the Centers for Disease Control and Prevention and the American Academy of
Orthopaedic Surgeons for the prevention of transmission of HIV or hepatitis C
to healthcare personnel were
followed22,23.
The types of prosthetic components that were utilized are shown in
Table II. Since 1987, all
components have been cemented. Antibiotic-loaded cement was not used.
Postoperative drains were used in all knees, and antibiotic prophylaxis was
administered for three to five days. No antithrombotic prophylaxis was
utilized.
Intensive physical therapy was started on the first postoperative day and
was continued twice daily while the patient was in the hospital. If the
postoperative range of flexion was <55°, knee manipulation was
performed between the seventh and fourteenth postoperative days, after the
wound had begun to consolidate. After discharge from the hospital, physical
therapy was continued five days a week for the first two to three weeks and
three days a week thereafter, usually for a period of six to nine weeks.
Of the sixty-eight patients (ninety knees) included in this series, thirty
(forty-one knees) had died at the time of this study, at a mean of 9.8 years
(range, 2.8 to 21.0 years) after the index surgery
(Table I). Thirteen patients
had died from complications of AIDS; three, from hepatic carcinoma secondary
to hepatitis-C infection; one, from acute liver failure; one, from
intracranial hemorrhage following a fall; and one, from a narcotics overdose.
The causes of the deaths of eleven patients were unknown. Five patients (six
knees) were lost to follow-up, at a mean of 4.7 years (range, 2.0 to 8.0
years) postoperatively. The available clinical information on patients who had
died or had been lost to follow-up was used for the analysis. Twenty-seven
patients (thirty-three knees) returned to our institution for a clinical
evaluation, including calculation of the Knee Society clinical and functional
scores24. Six
patients (ten knees) were evaluated with a telephone interview, and only the
Knee Society functional score and the pain section of the Knee Society
clinical score were obtained for those patients
(Table I).
Several methods of statistical analysis were performed with use of the
STATA program (StataCorp, College Station, Texas). Kaplan-Meier survivorship
analysis was performed to determine five and ten-year prosthetic survival
rates. The 95% confidence intervals were calculated with use of the method of
Dorey and Korn25.
Prosthetic failure was defined as removal of a component for any reason,
removal of a component or a diagnosis of infection (survivorship free of deep
infection at any point), and mechanical failure. A student t test, Fisher
exact test, or Wilcoxon rank-sum test was performed as appropriate to compare
differences between groups, with use of a two-tailed 0.05 level of
significance.
Early postoperative complications occurred in fourteen knees (16%).
The complications included an acute hemarthrosis in seven knees, heterotopic
ossification in two knees (in one patient who had had simultaneous bilateral
total knee replacement), three cases of Coombs-positive hemolysis, one wound
dehiscence at the time of a knee manipulation, and one nonfatal pulmonary
embolism.
The mean knee flexion contracture improved from 18° (range, 0° to
50°) preoperatively to 9° (range, 0° to 25°) in the early
postoperative period (p < 0.0001) and 8° (range, 0° to 30°) at
the time of the latest follow-up. The mean total flexion arc improved from
59° (range, 5° to 130°) preoperatively to 69° (range, 31°
to 115°) in the early postoperative period (p = 0.003) and 75° (range,
0° to 124°) at the time of the latest follow-up (p = 0.0002).
Postoperative knee manipulation was required for twenty-two knees, which had a
mean total flexion arc of 63° (range, 2° to 109°) at the latest
follow-up evaluation compared with 79° (range, 0° to 124°) for the
unmanipulated knees (p = 0.04).
Of the ninety knees, twelve (13%) required component removal. One required
removal because of aseptic loosening of a tibial component in a patient in
whom a tibial pseudotumor had developed; one, because of a supracondylar
femoral fracture that occurred three years after the index operation; one,
because of recurrent hemarthrosis secondary to synovial impingement between
the patellar and femoral components in a knee in which a metal-backed patellar
component had been originally used; one, because of a fracture of the
polyethylene insert; and nine, because of infection. Three of the aseptic
failures were treated with revision arthroplasty. The fourth aseptic failure
occurred in a fifty-six-year-old patient in whom the polyethylene insert
fractured eleven years after the index operation. At the time of the study,
the patient was awaiting revision surgery, and the result was considered to be
a failure for the survivorship analysis.
Of the ninety total knee replacements, fourteen (16%) in ten patients were
complicated by infection. Infection developed in association with ten (22%) of
the forty-six total knee replacements that were in patients with a bilateral
replacement, and both knees became infected in four of those patients. The
prevalence of infection associated with bilateral total knee replacement (22%)
was higher than that associated with unilateral total knee replacement (9%) (p
= 0.08). There were no early postoperative infections (less than six months
after the surgery) (Fig. 1).
Infection was diagnosed at a mean of 5.2 years (range, 6.3 months to 25.6
years) postoperatively. Staphylococcus aureus was grown on culture of
specimens from seven knees; Staphylococcus epidermidis, from three
knees; Streptococcus D, from one knee; alpha-hemolytic Streptococcus, from one
knee; and Citrobacter koser, from one knee. One infection was treated
elsewhere, and the infecting microorganism was unknown.
The infection was successfully managed with irrigation and
débridement as well as antibiotic therapy, without requiring component
removal, in five knees, whereas component removal was performed to treat the
infection in nine knees. Two knees in one patient, which became infected 2.3
years after the primary operation, were initially treated with irrigation and
débridement but the infection persisted, requiring removal of the
prosthetic components and application of antibiotic-loaded cement spacers. Of
the seven other knees that under-went component removal, two required
arthrodesis and five had a staged revision. No recurrent infections were
observed in any of the nine knees.
The prevalence of infection in the HIV-negative patients was similar to
that in the HIV-positive patients. Infection developed after four (13%) of the
thirty total knee replacements performed in HIV-negative patients and after
ten (17%) of the sixty total knee replacements performed in HIV-positive
patients (95% confidence interval for the difference, 0% to 10%; p = 0.5). The
mean CD4 count at the time of the surgery in the eight HIV-positive patients
in whom an infection developed was similar to that in the thirty-three
HIV-positive patients in whom an infection did not develop (434.4
cells/mm3 compared with 450.4 cells/mm3; 95% confidence
interval for the difference, 0 to 505 cells/mm3; p = 0.7).
The five and ten-year rates of survival of the total knee replacements,
with component removal for any reason (including one pending revision) as the
end point, were 91% (95% confidence interval, 84% to 95%) and 83% (95%
confidence interval, 72% to 90%), respectively
(Fig. 2). When component
removal for any reason or infection (survivorship free of deep infection at
any point) was considered to be the end point, the five and ten-year survival
rates were 84% (95% confidence interval, 76% to 90%) and 77% (95% confidence
interval, 65% to 85%), respectively (Fig.
2). When component removal for mechanical failure was considered
to be the end point, the five and ten-year survival rates were 96% (95%
confidence interval, 91% to 99%) and 96% (95% confidence interval, 90% to
99%), respectively (Fig.
2).
The Knee Society clinical score was available after thirty total knee
replacements, at a mean of 7.4 years (range, 2.0 to 21.3 years)
postoperatively. The mean clinical score was 81.3 points (range, 50 to 99
points) (Table III), with a
score of =85 points after sixteen total knee replacements (53%), between 70
and 84 points after eight (27%), between 60 and 69 points after four (13%),
and <60 points after two (7%). A pain score was obtained for thirty-eight
knees, and pain was never classified as worse than mild or occasional. Limited
motion was the principal reason for the decrease in the clinical score.
The Knee Society functional score was available after thirty-eight total
knee replacements, at a mean of 8.4 years (range, 2.0 to 21.6 years)
postoperatively. The mean functional score was 88.7 points (range, 15 to 100
points) (Table III), with a
score of =85 points after twenty-five total knee replacements (66%),
between 70 and 84 points after twelve (32%), and <60 points after only one
(3%). Most of the patients were able to walk unlimited distances, despite
restrictions in the range of motion. The score of <60 points was given to a
patient who was awaiting revision surgery for a fractured polyethylene insert
and who had a very limited range of motion and required the use of walking
aids.
The reported clinical results of total knee replacement in patients
with hemophilia have varied
considerably11,12,15,16,18,
with the prevalence of infection ranging from 0% to
17%11,12,14-16,18
and a rate of prosthetic survival of 90% after five
years14. This
variability in results could be explained by studies with small numbers of
patients, short to intermediate follow-up, and data from multiple surgeons and
centers.
The life expectancy of patients with hemophilia is less than that of the
general population, principally as a result of HIV infection. The authors of a
study of mortality of hemophilic men from 1993 to 1995 reported 40.4 deaths
per 1000 patient/years, 65% of which were HIV
related26.
Late infection is the main concern following joint arthroplasties in
hemophilic patients. In the present study, the prevalence of infection was
16%. Although that prevalence is consistent with the results of several
previously reported studies of total knee replacements in hemophilic
patients10,11,14,15,
it is considerably higher than the prevalence of 1% to 2% observed after total
knee replacements in the non-hemophilic
population27-29.
The high prevalence of late infection after arthroplasties in hemophilic
patients may be a consequence of an array of factors, including, but not
limited to, HIV status, hepatitis-C status, hematogenous infections from
frequent intravenous self-administration of clotting factor concentrate, or
other remote infection sources. A hematogenous source is suggested by the high
prevalence of infection observed with bilateral total knee replacement as well
as the typical skin microorganisms responsible for most of the observed
infections.
One suggested explanation for the higher prevalence of infection after
orthopaedic procedures in hemophilic patients is a higher prevalence of HIV
infections, with consequent immune
suppression21,30.
A higher prevalence of early infections in HIV-positive hemophilic patients
with CD4 counts of <200 cells/mm3 has also been
suggested21. Our
study, in which no early infections developed, does not support these
hypotheses. In our study, the prevalences of infection were similar in
HIV-negative and HIV-positive patients and the mean CD4 count at the time of
the surgery in the HIV-positive patients in whom an infection developed was
not considerably different from that in the HIV-positive patients in whom an
infection did not develop. Similar results were observed by Norian et
al.14.
It is unclear whether the use of antibiotic-loaded cement could be of
benefit in primary total knee replacements in hemophilic patients.
Antibiotic-loaded cement was not used in our series, and even though no early
periprosthetic infections (those developing less than six months
postoperatively) were observed, three infections developed within the first
year after the operation.
Infection was successfully controlled in five of the seven knees that were
managed with irrigation and débridement as well as antibiotic therapy.
In those five knees, the prosthetic components were well fixed and the
infection was caused by antibiotic-sensitive gram-positive microorganisms.
Those two factors have been associated with success after irrigation and
débridement for infections around total knee
replacements31. Of
the nine knees in which the infection was treated with component removal, two
received antibiotic-loaded cement spacers, two were successfully managed with
arthrodesis, and five had a staged revision. No recurrent infections were
observed in those nine knees.
As anticipated, the Knee Society clinical scores indicated
less-than-optimal knee motion because of arthrofibrosis, but they also
demonstrated excellent pain control because none of the patients rated the
pain as worse than mild or occasional. Despite the restricted range of motion,
97% of the patients for whom a Knee Society functional score was available had
reported an excellent or good functional result. The slight improvements in
motion associated with important reductions in articular pain produced
clinical results that were quite acceptable to the patients and that improved
their quality of life.
The survival rates of total knee replacements (with revision for any reason
as the end point) have been reported to be 91% and higher in nonhemophilic
patients32-38.
Despite the substantial anatomical challenges in this population, the present
study indicates that, on a mechanical basis, the rates of survival of total
knee replacements in hemophilic patients (96% at ten years) are equivalent to
those in nonhemophilic patients. However, the overall prosthetic survival rate
is less favorable than that in the general (nonhemophilic) population because
of infection. With survival free of infection at any point as the end point,
the five-year survival rate in our hemophilic population was 84%. Some of
those patients had early infections that were successfully treated with
irrigation and débridement, without implant removal. Furthermore, the
ten-year rate of implant survival free of infection was only 77%. The survival
rates observed in the present study are similar to those described by Norian
et al. in a relatively large hemophilic
population14.
Prevention of late infection would substantially improve the long-term
outcome of total knee replacement in the hemophilic population. In the hope of
reducing the prevalence of late infection, we have instructed patients
regarding the importance of meticulous antisepsis with self-infusion, regular
medical check-ups, immediate reporting of any signs of infection, and the use
of prophylactic antibiotics prior to dental work and any other procedures with
a potential for contamination. The use of short-term prophylactic antibiotics
prior to self-infusion should be considered. It is important to clearly inform
the patient of the higher risks associated with total knee replacement,
emphasizing the importance of meticulous attention to infection-prevention
recommendations postoperatively. The improvements in quality of life after
total knee replacement in hemophilic patients are substantial, but they must
be weighed against the risks of the procedure. ?