The literature on the treatment of ancient tuberculous arthritis with
late total joint replacement is sparse, and there is no clearly
recommended plan of care. We report the case of a patient who had
reactivation of Mycobacterium tuberculosis following
total hip arthroplasty after sixty-four years of quiescence. Several
key points regarding clinical strategies are explored, including
assessment of the need for chemoprophylaxis, the importance of long
periods of quiescence of active tuberculosis, and the options for
treating tuberculous infection at the site of a joint replacement.
The patient was a seventy-one-year-old man with a history of isolated
tuberculous arthritis of the left hip at the age of three years.
Initially, he had been treated with a hip-spica cast, followed by
an arthrodesis in 1932 at the age of five years. He received no
chemotherapy at that time as antituberculous drugs were not available.
The hip remained functional and pain-free for the next sixty-three
years, without evidence of tuberculous infection. When he was first
seen at our clinic in January 1997, he reported that he had had
progressive loss of flexibility, pain in the low back, and shortening
of the left lower extremity for two to three years. He noted increased
difficulty in performing foot hygiene and in maneuvering in constrained
spaces. He was otherwise in good health, and no other foci of tuberculous
infection were identified.
Physical examination revealed a moderately obese man with a short-leg
gait and a shortened swing phase on the left. The left hip was fused
in 45° of flexion, 10° of abduction, and 15° of external rotation.
The left lower extremity had 3 cm of apparent shortening. There
was palpable contraction of the gluteal muscles with single-leg
stance. Pain was localized to the lumbosacral region with forward
flexion of the back. The findings on examination were otherwise
unremarkable. The right lower extremity was normal.
Radiographs showed a mature fusion of the left hip (Fig. 1) and a normal
ipsilateral knee. Pelvic radiographs also demonstrated hypertrophic
lumbar degenerative disc disease. An electromyogram revealed a normally
functioning gluteus maximus and no voluntary motor potentials in
the gluteus medius or the gluteus minimus. The erythrocyte sedimentation
rate was 17 mm in the first hour. The white-blood-cell count was
7000 mm3 (7.0 109/L).
A chest radiograph showed a 1-cm smooth nodule in the left upper
lobe, which was thought to be most consistent with granuloma. A
PPD (purified protein derivative [tuberculin]) skin
test was positive.
On the basis of the progressive loss of function, the pain in
the low back, and the lack of evidence of active disease, the patient
was considered a candidate for either revision osteotomy to a more
acceptable position of arthrodesis or conversion to a total hip
arthroplasty. The patient opted for the latter. After discussion
with infectious disease consultants, we elected to defer specific
mycobacterial chemoprophylaxis.
The arthroplasty was performed in March 1997. Both the femoral
component and the acetabular component were inserted without cement.
The reconstruction resulted in lengthening of approximately 2 cm,
leaving a net apparent residual shortening of 1 cm. The abductor
muscles were found to be attenuated. The posterior portion of the
gluteus maximus and the posterior fascia lata were transferred to
the greater trochanter to augment abductor muscle function. All
samples of synovial fluid and tissue were negative for organisms
on gram stain and acid-fast stain, and there was no sign of acute
inflammation or granuloma on frozen-section analysis. All cultures
of tissue (including those for mycobacterium) also were negative.
Five months postoperatively, the patient returned to our clinic because
of a painless, diffuse, erythematous swelling about the proximal
aspect of the left thigh with serosanguineous drainage from the
distal portion of the incision. There was no purulent discharge
or odor. The radiographic findings were unchanged from those that
had been noted immediately postoperatively (Fig. 2). The patient
was taken to the operating room, where a large collection of fluid
was found extending deep to the fascia and down into the joint;
exuberant synovitis was evident. The components were secure. The
joint was dislocated, and the components were scrubbed with Betadine
(povidone-iodine). The polyethylene liner was replaced, and a synovectomy
was performed. Histologic examination revealed chronic inflammation
with fibrosis, focal granulation tissue with acute inflammation,
and numerous multinucleated giant cells associated with caseating
granuloma (Fig. 3).
No organisms were identified on gram stain or acid-fast smear of
intraoperatively obtained fluid and tissue. Standard cultures showed
no growth. Cultures for Mycobacterium tuberculosis were
positive at fifteen days for organisms sensitive to all antituberculous
drugs except streptomycin. The patient was initially treated with
isoniazid, rifampin, ethambutol, and pyrazinamide after consultation
with the state department of health. The use of pyrazinamide was
later discontinued when sensitivities were finalized. This regimen
was continued for one year postoperatively.
At the most recent follow-up examination in November 1999, twenty-six
months following débridement and fourteen months following
cessation of antituberculous therapy, the patient was pain-free
and without clinical or radiographic evidence of infection in the
hip. He continued to use a 0.5-in (1.27-cm) heel lift in the left
shoe to correct for leg-length discrepancy. He had improved abductor
strength (grade 4 of 5) and was able to walk without assistive devices.
The range of motion of the left hip was 80° of flexion, 30° of abduction, 30°
of internal rotation, and 15° of external rotation.
Tuberculous arthritis accounts for approximately 1% to
5% of all tuberculous infections. Ninety percent of cases
are monarticular, with the hip and knee being most commonly affected1. Although tuberculosis is less prevalent
in the United States than it was fifty years ago, the emergence
of the human immunodeficiency virus, an aging population, the large
number of chronically immunosuppressed people, the ease and frequency
of worldwide travel, and multidrug-resistant strains of tuberculosis
have produced a resurgence in clinically apparent tuberculosis and
a renewed interest in its diagnosis and treatment.
Chemotherapy alone can be used to treat very early tuberculous
arthritis, with good outcomes; however, the disease is rarely diagnosed
at an early stage. The conventional treatment for more advanced
disease has long been arthrodesis2.
Arthrodesis typically relieves pain and provides a more stable,
functional joint; moreover, it has been thought that a solid osseous
fusion provides local inhibition of the infectious process2,3. The goal is to produce osseous
ankylosis rather than allowing the natural course of the disease
to result in fibrous ankylosis and a chronically painful joint.
Chen and Lee found that arthrodesis led to a solid osseous fusion
in thirty-two (86%) of thirty-seven patients who had tuberculous
arthritis2. Patients who presented
before the advent of antituberculous chemotherapy, which was introduced
in 1945, were treated by surgical or natural arthrodesis alone2. In modern times, arthrodesis has
been combined with chemotherapy.
Total joint arthroplasty has become an accepted alternative treatment
for hips that are ankylosed as a result of tuberculous arthritis.
Indications for conversion to total hip arthroplasty include progressive
worsening of pain in the low back or the contralateral hip (with
a decrease in function and mobility), and malposition of the hip
following fusion4. Numerous authors
have reported substantial improvement in function, mobility, and
pain scores following total hip arthroplasty3,5,6.
Y.Y. Kim et al. reported an increase in the mean rating for function,
from 62 points preoperatively to 85 points following total hip arthroplasty,
in a study of thirty-eight patients (forty-four hips) who had tuberculous
coxarthrosis7. Laforgia et al.,
in a series of seventy-two patients with prior tuberculous infection
of the hip, reported the results of total hip arthroplasty in a
subset of thirty-four patients with a spontaneously fused or arthrodesed
hip6. At a mean of five years
after the arthroplasty, the mean score for motion, according to
the numerical grading system of Merle d’Aubigné and
Postel, had improved from 1.0 point preoperatively to 4.5 points;
the mean score for function had improved from 3.6 points to 4.7
points; and the mean score for pain had improved slightly, from
5.6 points to 5.7 points. The results in patients with previous
tuberculous infection are generally not as dramatic as those seen
after primary total hip arthroplasty for other diagnoses, mainly
because of the presence of residual fibrosis in the previously infected
hip3.
There is clearly a danger of reactivation of tuberculosis after total
hip arthroplasty. Although there is no substantial evidence regarding
the exact role of chemoprophylaxis, Berbari et al., in a series
of 166 patients who had had a total hip arthroplasty because of
tuberculosis, reported that eight (9%) of ninety-two patients
who had received some antituberculous prophylaxis had a reactivation
of the infection compared with fourteen (19%) of seventy-four
who had not8. The duration of
the prophylactic therapy, the choice of drug, and the selection
criteria for the patients who received prophylaxis varied widely.
Jupiter et al. reported no recurrence of infection in a series of
seven patients who had been treated with one antituberculous drug
for one year prior to total hip arthroplasty9.
Y.H. Kim et al., in a series of forty-four total hip arthroplasties
in patients with old tuberculosis, reported reactivation of infection
in six (14%) despite the three months of chemoprophylaxis
that had been given preoperatively to patients with laboratory evidence
of active infection3. Y.Y. Kim
et al., in a series of twenty total hip arthroplasties, reported
reactivation of infection in five (25%) despite administration
of rifampicin, isoniazid, and ethambutol three weeks prior to and
six to nine months after total hip arthroplasty7.
Conversely, other authors have reported success without chemoprophylaxis.
Hardinge et al. reported no recurrences of infection after total
hip arthroplasty in twenty-one patients who had not received prophylaxis10. On the basis of the overall trend,
several authors have empirically recommended chemoprophylaxis prior
to total hip arthroplasty, especially in patients who had not been
treated with chemotherapeutic agents primarily, in those with evidence
or suspicion of active infection, or in those with a short (less
than ten-year) period of quiescence3,5,6,9.
Treatment of patients with tuberculous infection after total
hip arthroplasty presents a major and infrequent challenge. Most authors
have employed a combination of surgical and medical treatments.
Of a total of twenty-one cases of reactivation of infection that
were identified in our review of the literature3,6-8,11-13,
twelve (57%) were treated with antibiotics and removal
of the components (followed by staged revision in some); four (19%)
were treated with antibiotics, aggressive débridement, and
retention of the components; and five (24%) were treated successfully
with antibiotics alone. The mean duration of follow-up for these
twenty-one patients was 2.7 years. Multiple treatment modalities
are described in the literature, and with so few reported cases
it is not possible to convincingly demonstrate the superiority of
any one regimen. Antibiotic regimens employ multiple drugs on the
basis of the sensitivities of the infecting organisms and their
propensity to rapidly develop resistance to single-drug therapy.
The Centers for Disease Control recommends that tuberculous arthritis
be treated initially with isoniazid, rifampin, pyrazinamide, and either
ethambutol or streptomycin if isoniazid resistance is suspected.
Therapy should continue for at least nine months, with conversion
to isoniazid and rifampin if the organism grown on culture is isoniazid-sensitive14.
Clearly, there is no period of quiescence beyond which the possibility
of reactivation is eliminated. Our patient had the longest period
of quiescence (sixty-four years) followed by reactivation that has
been reported, to our knowledge. Y.H. Kim et al. suggested that
a longer duration of quiescence may decrease the likelihood of recurrence3. In their series, none of twenty-three
infections that had been quiescent for more than ten years recurred,
whereas six of fifteen that had been quiescent less than ten years
recurred. Hardinge et al., in a series of twenty-one infections
that had been quiescent for at least twenty years prior to total
hip arthroplasty, reported no reactivation despite the absence of
prophylaxis10.
On the basis of both the current case and the inconsistent reports
in the literature, we concluded that one cannot be too diligent
in weighing the risks and benefits of prolonged antituberculous
prophylaxis before and after total hip arthroplasty. In the absence
of confounding medical conditions and contraindications to specific
drug therapy, serious consideration should be given to extended
perioperative antituberculous therapy, irrespective of the duration
of clinical quiescence.
Note: The authors thank Elizabeth Frauenhoffer, MD, for providing
pathology slides and descriptions for this case report.