The study population consisted of all patients with a fungal
infection of the spine treated by the authors from 1973 to 1989
in three major teaching institutions: the University of Miami School
of Medicine, the Case Western Reserve University School of Medicine,
and the George Washington University School of Medicine. Eleven
patients were treated for osteomyelitis of the spine caused by a
true fungus. The series of patients was sequential and inclusive.
The classification system for neural deficits in patients with
spinal osteomyelitis, as developed by Eismont et al.16, was used in this patient population.
With this system, sensory deficits are ignored for the purpose of analysis
and neural deficits are categorized according to the degree of motor
paralysis. Class A is distal strength of 0 or 1 (of 5), class B
is distal strength of 2 or 3, class C is distal strength of 4, and
class D is distal strength of 5 (normal).
The medical records and roentgenograms were available for every
patient. There were eight male and three female patients ranging
in age from twenty-seven to seventy-eight years (average age, fifty-five
years). Long-term follow-up of nine surviving patients was performed
by direct examination by the authors or by obtaining information from
the patient’s private physician (one patient). One patient
was contacted by telephone to augment the information. Patients
were specifically asked about any spinal pain and were examined
for persistent paralysis or symptoms of recurrent infection. Of
the eleven patients, one died of generalized sepsis thirty-three
days after the diagnosis and another died of a gastrointestinal
hemorrhage at five months. The duration of follow-up of the nine surviving
patients ranged from one year and six months to sixteen years and
four months, with an average of 6.3 years. The patient who was followed for
one year and six months was subsequently lost to follow-up.
Of the eleven patients, ten presented with severe unremitting
pain localized to the level of the infection; two of the ten also
had radicular leg pain. Nine patients had paresthesias and/or
dysesthesias in the legs. Nine patients presented with motor weakness, and
in the tenth patient weakness developed during the hospital stay.
On presentation, according to the classification system of Eismont
et al.16, one patient had class-A
paralysis, two patients had class-B, six had class-C, and two had
no paralysis (Class D).
All but two patients had a medical condition or were undergoing
treatment that is associated with compromise of the immune system:
four patients had received high-dose systemic corticosteroids, one
had received repeated local corticosteroid injections, two had diabetes
mellitus, one was undergoing chemotherapy for a tumor, and one was severely
malnourished (Table I).
All of the patients had a long delay between the onset of symptoms
and the administration of an antifungal agent. Ten patients had
an average delay in the diagnosis of ninety-nine days (range, seven
to 365 days), and the eleventh patient reported a history of a draining
sinus in her back for nine years prior to treatment.
Of the eleven patients, only one presented with a fever of more
than 101°F (38°C), and four were afebrile upon presentation17. The erythrocyte sedimentation rate
was elevated in ten patients; it was <51 mm/hr
in three of these patients and 51 mm/hr in seven (range,
14 to 120 mm/hr in the ten patients). The white blood-cell count
was elevated in only three of the eleven patients; it was <20,000
cells/mm3 (<20.0 × ¥ 109/L)
in two of them and greater than that value in one.
Roentgenographic Analysis
On presentation, roentgenograms demonstrated peridiscal erosions
and decreased disc-space height in ten patients. In the eleventh
patient, who had the most chronic infection, there was diffuse sclerosis involving
multiple thoracic and lumbar vertebrae with lateral vertebral body
scalloping. The levels of the spinal infections are shown in Table II.
Technetium-99m bone scans and gallium scans18 were
done in all patients and were found to be false-negative in two.
In three patients, trispiral polytomography was used prior to the
advent of computerized tomography, and it was useful for identifying
the end-plate erosions. Computed tomography and magnetic resonance
imaging in the more recently seen patients were found to be uniformly
useful for delineating extrinsic cord compression, paraspinal abscess,
and the extent of adjacent disc-space or vertebral body involvement19,20.
Mycologic Analysis
Tissue obtained with Craig needle biopsy, material obtained with
fine-needle aspiration, or operative specimens revealed the pathogenic
fungus in all eleven patients. The organism was identified with
a positive fungal culture, potassium hydroxide slide preparation,
or Gomori methenamine silver stain21-23.
The types of fungi isolated from the spine are listed in Table II.
Two patients had positive blood cultures. In one patient, fungus
grew from the tip of a central venous catheter. Two patients had
identical fungi grown on culture of both urine and spinal biopsy specimens.
One patient (Case 2) with disseminated aspergillosis had the organism
identified in the mitral valve, eye, superior mesenteric artery,
femoral arteries, and spine. This patient also had Nocardia
asteroides sepsis and Candida albicans sepsis.
Cerebrospinal fluid complement fixation titers indicated coccidioidal
meningitis with coexistent spinal osteomyelitis in one patient,
with changes in mental status, psychosis, and severe headaches24.
Postoperative fungal osteomyelitis developed following elective
spine surgery in three patients, hematogenous spread from a distant
site occurred in four, and local extension from an adjacent area
of fungal infection occurred in two. The source of the fungal infections
in the other two patients was unknown. However, one of these patients
was impaled through the third lumbar level with a steel rod in a
motorcycle accident, and direct contamination was the presumed source
of infection.
Treatment
Ten patients were treated with intravenous amphotericin B, with
the total dose ranging from 1.0 to 3.35 g, and one also received
intrathecal amphotericin B for the treatment of coexistent coccidioidal meningitis25. Intravenous miconazole and oral
ketoconazole were used to treat a Petriellidium boydii infection
in the one remaining patient because this fungus has exhibited resistance
to amphotericin B26. Four patients
received more than one adjunctive antifungal agent, including rifampicin,
5-fluorocytosine, and ketoconazole (Table II)11,26.
All patients had at least one minor complication from the amphotericin
B, including nausea, vomiting, phlebitis, renal dysfunction, neutropenia,
anemia, and thrombocytopenia. The ketoconazole therapy caused severe
pruritus, which was relieved by antihistamines, allowing a complete
therapeutic course of 44.8 g of ketoconazole to be given.
Ten patients were treated with surgical débridement of
the infected area of the spine. The indications for operative intervention
were the need to (1) obtain tissue for diagnosis, (2) debride the
spine and drain an abscess, (3) decompress the neural elements,
and (4) treat a case that was refractory to medical treatment27. Five patients had an initial posterior
approach with laminectomy, and three of these patients later required
an anterior approach with disc excision, partial or complete corpectomy,
and strut-grafting with either an autogenous tricortical iliac crest
graft or rib grafts. The indications for additional anterior surgical
débridement and stabilization were progressive osseous
destruction and instability (Case 1), an acute Brown-Séquard
syndrome after spinal stabilization with Luque rods and segmental
wires (Case 5), and persistent severe back pain and nonunion despite
amphotericin-B therapy (Case 10). In a fourth patient (Case 9),
the lumbar spine became unstable following a wide laminectomy and
bilateral facetectomy, with 50% retrolisthesis of the first on
the second lumbar vertebra; the patient died five months postoperatively.
The spine was still unstable, and surgical treatment through a posterior approach
in this patient should be considered a failure. In contrast, five
patients were treated primarily with anterior débridement
and stabilization, and none required additional surgery. Of the
ten patients who were operated on, nine had an arthrodesis: seven
had it anteriorly and two, posteriorly. Three patients received
posterior instrumentation.
A sagittal deformity of >20° did not develop after any
of the spinal arthrodeses. All patients had a successful fusion,
except one (Case 2), who died thirty-three days postoperatively.
In three patients, the fungal infection occurred following a
discectomy (Cases 3, 10, and 11). One patient (Case 11) who had
a postoperative disc-space infection with Candida albicans after
multiple laminectomies and disc excisions and at least fifteen local
corticosteroid injections in the year preceding the onset of infection
was successfully treated nonoperatively with a pantaloon spica cast
for three months and intravenous amphotericin B.
Four patients had a delay in treatment of less than sixty days,
and three of them had a motor deficit upon presentation; two recovered
to class D (normal). In comparison, seven patients had a delay in treatment
of greater than sixty days, and six of them had a motor deficit;
none fully recovered to class D.
The neural status was known at the time of final follow-up for
the eleven patients. The patient who had presented with Eismont
class-A paralysis secondary to a traumatic impalement during a motorcycle accident
still had class-A paralysis at the time of final follow-up28. Of the two patients with class-B
paralysis, one had no change and one had improvement to class-C.
Of the six patients with class-C paralysis, four had no change and
two had improvement to class-D. Of the two patients with class-D
(no) paralysis, one remained neurologically normal and one (Case
5) lost function (to class-B) following the original surgery but
had recovery to class-D at the time of final follow-up.
Of the nine surviving patients, two had severe mechanical low-back
pain requiring narcotics and two others had mild mechanical back
pain. All four had had a low lumbar spinal infection.
Complications
Of the eleven patients, nine had at least one major complication
during their hospitalization. Two patients died in this series.
One patient with disseminated aspergillosis died thirty-three days
after multiple-level anterior thoracic and lumbar corpectomies and
grafting with autogenous rib strut grafts. His neurologic status
had remained at class C. The second death occurred five months after
laminectomy and disc biopsy for Candida tropicalis osteomyelitis
(Case 9) and was a result of massive upper gastrointestinal hemorrhage.
The patient was being treated for ulcerative colitis with high-dose systemic
corticosteroids and was undergoing renal dialysis. His neural status
had also remained at class C.
In reviewing these eleven cases of fungal infection of the spine
treated over a sixteen-year period, we identified several trends
that affect diagnosis and treatment.
A delay in treatment was common because of the difficulty in
diagnosis, as has been reported in other series5,27,29,30.
Besides being uncommon pathogens (two of sixty-five cases of spondylitis
were caused by a fungal organism in the study by Eismont et al.16), fungal organisms are slow-growing
and difficult to identify by culture21.
However, several factors are predisposing more patients to fungal
infection5,27. They include immunodeficiency
secondary to infection with the human immunodeficiency virus, the
widespread and injudicious use of broad-spectrum antibiotics, the
use of corticosteroids and other immunosuppressive drugs, and the
use of parenteral hyperalimentation31.
In our series, seven of the eleven infections were due to Candida
alone. The results therefore might be more reflective of this class
of organism than of general fungal infections of the spine.
The time prior to treatment affected the outcome in these patients.
Longer delays in the initiation of treatment were associated with
a less favorable neural outcome, and it is likely that these long
delays were an important reason why seven of the eleven patients
did not have full neural recovery. This suggestion is in agreement
with the observations of Kushwaha et al., who noted that, although
many of their twenty-five patients with coccidioidomycosis did well
despite a delay in treatment, such a delay often complicated treatment5.
Because our study was a retrospective review of the experiences
of three spine surgeons who treated a variety of fungal infections,
the indications for operative treatment cannot be accurately defined. However,
the fact that ten of the eleven patients underwent spinal surgery
suggests that medical management alone may not produce acceptable results
in patients with fungal spondylitis5.
Four of five patients who had initial spinal débridement
from a posterior approach by means of a laminectomy were considered
to have had a failure of treatment. Three required additional débridement, decompression,
or stabilization from an anterior approach, and the fourth died
before that was possible. In their study of patients with pyogenic
and fungal spinal osteomyelitis, Eismont et al. concluded that anterior
decompression and stabilization yielded predictably better results
than did laminectomy in patients with spinal cord compression16. In our review, it appeared that,
even without neural compression, the results were better when the
surgical approach to the spine was anterior. With the infection
destroying the anterior and middle columns of the vertebrae, a laminectomy
removing the posterior elements will only produce instability of the
spine16,32.
Once treatment of the infection has been completed, the long-term
prognosis for neural recovery and pain relief remains uncertain.
At the time of follow-up, only four of our nine surviving patients were
neurologically normal. Additionally, four patients had persistent
low-back pain and two of them were functionally disabled by the
pain. Fortunately, once full medical and surgical treatment was completed,
no patient in this series had a recurrence of the infection.
We presented a retrospective review of the treatment of fungal
spondylitis. In most of the cases, there was preexisting immunocompromise,
a delay in diagnosis, difficulty in treatment, and complications
associated with the antifungal medication. The long-term clinical
outcome did not seem to be related to the specific species of fungus
but rather to the time between the onset of the symptoms and the
treatment of the infection. Delay in diagnosis led to poorer results
in terms of final neural recovery. When tissue is obtained for the
diagnosis of suspected spinal infection, fungal cultures should be
done. We recommend presenting patients with a guarded prognosis
and informing them of the many possible complications associated
with treating fungal spondylitis.