Downloaded from www.ajronline.org by 117.30.56.203 on 10/14/15 from IP address 117.30.56.203. Copyright ARRS. For personal use only; all rights reserved
527
Diagnosis of Osteomyelitis Foot in Diabetic Patients: 1111n-Leukocyte
George Larcos1 Manuel L. Brown1’2 Ross T. Sutton3
The noninvasive currently available
of the Value of
Scintigraphy
diagnosis radiologic
of the foot in diabetic patients with imaging techniques is often difficult.
of osteomyelitis and radionuclide
Recently, 111In-labeled leukocyte scintigraphy has been proposed as an attractive alternative. Accordingly, we retrospectively reviewed 51 “1ln-labeled leukocyte scans, 49 technetium-99m bone scans, and 49 plain radiographs obtained in 51 adults with diabetes in whom osteomyelitis of the foot was suspected. The sensitivity and specificity of these
techniques
were
evaluated
in all patients,
as well
as in a subgroup
of 11
patients with neuroarthropathy. Results with “11n-labeled leukocyte scans were also examined in subsets of patients with soft-tissue ulcers (n = 35) and those receiving antibiotics during investigation (n = 20). Confirmation or exclusion of osteomyelitis was made surgically in 28 patients and clinically in 23. Fourteen patients had osteomyelitis. Bone scans were most sensitive (93%) but least specific (43%); plain radiographs were most specific (83%) but least sensitive (43%). 111In-labeled leukocyte scans were both
sensitive
(79%)
and specific (78%),
and remained
useful in patients with neuroarthrop-
athy, soft-tissue ulcers, and antibiotic treatment. Poor spatial resolution contributed to the false-negative and false-positive 111ln-labeled leukocyte scans, suggesting that this technique should not be interpreted independent of other tests. “‘In-labeled leukocyte scans are a valuable diagnostic tool for the diagnosis of pedal osteomyelitis in diabetic patients. AJR 157:527-531,
September
1991
Diabetic patients in whom osteomyelitis of the foot is suspected pose a common yet formidable diagnostic problem [1 ]. Coexisting disorders such as peripheral vascular disease, cellulitis, and neunoarthropathy obscure the clinical and nadiologic manifestations of osteomyelitis [1 , 2]. An inaccurate diagnosis is associated with deleterious consequences [3]. Plain radiographs, 99mTcmethylene diphosphonate
(MDP) scans,
and 67Ga-citrate
scans are widely
considered
to have limited
diag-
nostic accuracy [1 , 3-5]. Recently, several investigators [6-1 4] reported favorable results in osteomyelitis when ln-labeled leukocytes were used. However, most of these series have contained few patients with diabetes mellitus, and the Received January 28, 1991; accepted after re vision April 17, 1991. 1 Section of Nuclear Medicine, Mayo Clinic and Mayo Foundation, Rochester, MN 55905. 2 address: Division of Nudear Medicine, Department of Radiology, University of Pittsburgh Medical Center, Desoto at O’Hara Sts., Pittsburgh, PA 15213. Address reprint requests toM. L. Brown.
of Diagnostic Radiology, Mayo Clinic and Mayo Foundation, Rochester, MN 55905. 3Department
0361-803X/91/1573-0527 © American
Roentgen
Ray Society
conclusions regarding the usefulness of 11n-WBC scanning in neuroarthropathy have been conflicting [4, 8, 9, 1 5]. Further, diabetics often have soft-tissue ulcers or are receiving antibiotics during investigation, yet there is a paucity of information regarding
the potential
WBC scintigraphy.
limitations
Initial evaluation
imposed
by these
of CT scanning
factors
on results
been promising, but both are more expensive than scintigraphy respective roles in imaging strategies are at present undefined [1]. The purpose of this study was to determine the usefulness
scintigraphy
in a large
heterogeneous
group
with
[1 6] and MR imaging
of diabetic
patients
[3] of
11n-
[1 7] have and
their
1111n-WBC
referred
for
investigation of possible pedal osteomyelitis. The value of this test was also assessed in subsets of patients with advanced neunoarthnopathy, current antibiotic therapy, and foot ulcers. Further, since 111InWBCs may localize falsely at sites of
528
LARCOS
hyperemia and the interpretative
noninfective inflammation criterion requiring uptake
Downloaded from www.ajronline.org by 117.30.56.203 on 10/14/15 from IP address 117.30.56.203. Copyright ARRS. For personal use only; all rights reserved
tionately greaten osteomyelitis.
Materials
the 5 years or
99’Tc-MDP
to
indicate
were
from April
1 985 to March diabetes
for suspected
selected
in whom
1 990, 76 adults
mellitus
underwent
pedal osteomyelitis.
a final diagnosis
1In-
Fifty-one
was proved
with
pa-
by either
surgery (bone biopsy or culture, or deep wound culture during dobridement) or clinical follow-up in the outpatient department. Accordingly,
51 1111n-WBC scans,
49 “Tc-MDP
scans,
and 49 plain radio-
graphs were reviewed. The studies of the remaining 25 patients were excluded from analysis because treatment was subsequently undertaken at other institutions, with no correspondence regarding outcome (n = 20), or these patients had multisystem disorders with inadequate
documentation
for the foot (n
= 5).
The study group included 31 men and 20 women with a mean age of 62 years (range, 30-88). The majority of patients (n = 49) had diabetes mellitus for at least 1 2 months; however, the remaining two subjects were diagnosed only 1 month before examination for osteo-
myelitis. The mean duration of diabetes was 14 years. Thirty-five patients had ulcers adjacent to suspected areas of osteomyelitis. Twenty patients were receiving antibiotics at the time of investigation; treatment
in 1 1 had been
discontinued
at least
2 days
before
or absence of osteomyelitis and by a clinical follow-up
months;
mean,
26 months)
the
was established by surgery of at least 2 months (range,
in the rest. Patients
who were
diagnosed by clinical means were considered not to have osteomyelitis if their signs and symptoms responded to topical therapy and/or a brief (less than 2 weeks) course of antibiotics with no subsequent relapse. Culture of sinus tract specimens was considered indetermi-
nate for osteomyelitis cultures
or biopsies
soft-tissue ulcers The 111InWBC
and was not used for diagnosis were
obtained
or infection scan was
only
after
surgical
[3]. obtained
[3]. Bone
debridement
September
at 24 hr by using one of a number
1991
of large-
7500 Orbiter, General Electric 400T or 500 Maxi), all equipped with a medium-energy collimator. Multiple views of the feet were acquired for 10 mm with symmetric 20% windows placed around the 173- and 247-keV photopeaks of 1 1 1ln. The three-phase bone scan was conducted after the IV injection of approximately 800 MBq (21 mCi) of 99mTc-MDP. The dynamic gamma
cameras
(Siemens
by using consecutive
5-sec/frame
images in the
plantar or anterior projection in order to visualize optimally the area of interest. Blood-pool and delayed 3-hr scans of both feet were obtained for 500,000 counts, each in multiple projections. All images were obtained by using the same gamma camera equipped with a
low-energy
all-purpose
collimator
and a
20%
window
centered
sym-
metrically over 1 40 keV. The 1111n-WBC and mTc-MDP scans were usually obtained on the same day (n = 47) and in all cases within 6 days of each other (n = 49). When the tests were performed on the same day, the 111ln-WBC scan was acquired before injection of S9mTo-MDp Foot radiographs were taken on average within 8 days of these scans (range, 0-58). The mean interval between the radionuclide studies and surgical correlation (when available) was 5 days (range, 0-41). The radionuclide scans and radiographs were analyzed retrospectively by two experienced observers who had no knowledge of the final diagnosis. Limited clinical information (e.g., site of suspected infection) was made available to the readers. The observers were
for osteomyelitis.
The
11n-WBC scan was considered abnormal if focal accumulation of radionuclide activity in bone exceeded background radioactivity. Uptake was determined to be in bone or soft tissue by comparison with the bone scan images. The three-phase bone scan was considered indicative of osteomyelitis according to previously described criteria 11
1ln-WBC study (mean, 7 days), whereas 20 patients were not started on antibiotics until after the completion of noninvasive tests.
2-50
Images were obtained field-of-view
asked to code studies as positive or negative
11
The presence in 28 patients
AJR:157,
AL.
phase was acquired
noninsulin-dependent
WBC scintigraphy tients
than
[8], we tested to be of propor-
and Methods
During insulin-
intensity
ET
of
with 111lnxine according to modification [18] of the method of Thakur et al. [19]. After preparation of the cells, 18 MBq (470 MCi) of labeled leukocytes was reinjected.
[20].
In
brief,
focal
arterial
uptake of radionuclide positive. Osteomyelitis
when (1 ) bone destruction soft-tissue ized
swelling
osteopenia
hyperemia
with
increased
was present alone or in combination
or osteopenia or
associated
by bone on delayed views was considered was diagnosed on the basis of radiographs
periosteal
or peniosteal reaction
reaction,
occurred
in the
with
or (2) localabsence
of
fracture or neuropathic joint disease. The presence of significant neuroarthropathy also was recorded, as this may influence the sensitivity and specificity of 1111n-WBC scans. Finally, two authors reviewed scans in all patients in whom uptake of 111In-WBC and “Tc-MDP occurred in identical areas of bone, in
order between
to determine infection
whether and
the relative
“reactive”
Fig. 1.-A, tion shows
uptake
hyperemia.
“Tc-MDP uptake
in right
This
could was
distinguish assessed
scan in plantar third
and fourth
projecmeta-
tarsophalangeal joints of a 47-year-old insulindependent diabetic patient with chronic plantar ulcer. B, Corresponding
1111n-WBC scan shows intense activity at this site. Osteomyelitis was confirmed
B
at surgery.
AJR:157,
September
TABLE
1: Sensitivity
“Tc-MDP
lN-WBC
1991
and Specificity
Scans and Radiographs
Osteomyelitis
in Diabetic
SCINTIGRAPHY
of 111ln-WBC
and
in the Diagnosis
of Pedal
Patients
No. of Pa tients (%)
Downloaded from www.ajronline.org by 117.30.56.203 on 10/14/15 from IP address 117.30.56.203. Copyright ARRS. For personal use only; all rights reserved
Group/Study Sensitivity
Specificity
11/14 (79) 13/14 (93) 6/14 (43)
29/37 (78) 15/35 (43) 29/35 (83)
1/1 (100) 1/1 (100)
7/10 (70)
All patients
ln-WBC “Tc-MDP Radiograph
2/10 8/10
(20) (80)
Radiograph Antibiotics
0/1
(0)
ln-WBC
4/5
(80)
11/15 (73)
11/13
(85)
17/22 (77)
Soft-tissue ulcers 1111n-WBC
=
qualitatively
=
true positive/(true positive + false negative); + false positive).
Note-Sensitivity true negative/(true
specificity
negative
by visually
of radionuclide
comparing
in diseased
for each agent
the maximal
bone relative to background
uptake
radioactivity.
Results Osteomyelitis
of the foot
was
diagnosed
Eleven of these 14 cases were identified
in 14 patients.
correctly
by using
1111n-WBC scanning (Fig. 1). In one patient with a falsenegative study, uptake of ln-WBC was noted in the foot but was considered to be in soft tissues. The radiograph and
bone scan surgery
were
a sinus
also unremarkable tract
leading
in this patient,
to an infected
left second
OSTEOMYELITIS
529
normal. At surgery, necrotic bone was nesected and cultured. In one case, no pathogenic microorganisms were grown; however, histologic examination showed changes consistent with chronic osteomyelitis (Fig. 2). Of the 37 patients without osteomyelitis, there were 29 true-negative and eight false-positive ln-WBC studies. Of the latter, surgical correlation was available in five. In three of
these, there were soft-tissue microabscesses ln-WBC scan had been incorrectly ascribed
Neuroarthropathy
ln-WBC Tc-MDP
IN PEDAL
but at
that on the
to bone; in the other two cases, no bone on soft-tissue infection was present. The mTc-MDP scan was most sensitive but least specific for osteomyelitis, whereas nadiographs were specific but insensitive (Table 1). Eleven patients had neuropathic joint disease on radiographs. The 1ln-WBC scan was both sensitive and relatively specific for osteomyelitis in this group (Fig. 3). However, the 99Tc-MDP scans and radiographs lacked specificity and sensitivity, respectively. 11n-WBC scintigraphy was also sensitive and specific in patients with soft-tissue ulcers and in those subjects receiving antibiotics during investigation (Table 1). The results for 1 3 in whom 1ln-WBC and 99mTcMDP uptake occurred concordantly in bone were analyzed separately. Uptake 1ln-WBCs was proportionately greaten than uptake of 99mTc-MDP in eight of nine patients in whom osteomyelitis was confirmed (89% sensitivity). However, two of four patients without osteomyelitis also had this feature (50% specificity).
Discussion
meta-
tarsal head was identified. In the other two patients, the bone scan was positive at sites of chronic infection (left ankle and great toe, respectively), whereas the 11n-WBC scans were
Pedal osteomyelitis is a common complication of diabetes, occurring in up to 1 5% of patients [3]. Early detection reduces the likelihood of chronic osteomyelitis [3]. However, owing to
5#’
11
:a.
L
J:
.\....‘
::
B
o
t4kLA
Fig. 2.-65-year-old man with chronic pain and cellulitis of left great toe. Plain radiograph (not shown) revealed lytic changes in distal aspect of proximal first phalanx. This was considered indicative of osteomyelitis. A and B, “Tc-MDP blood pool (A) and delayed (B) plantar views were interpreted as positive for osteomyelitis. C, Corresponding ‘“ln-WBC scan shows no abnormality. However, histologic examination of resected necrotic bone showed changes consistent with chronic osteomyelitis, indicating that 111ln-WBC scan was falsely negative.
530
LARCOS
,:1:: Downloaded from www.ajronline.org by 117.30.56.203 on 10/14/15 from IP address 117.30.56.203. Copyright ARRS. For personal use only; all rights reserved
,
ET
AJR:157,
AL.
.
..
..
1
:
;
,
‘
#{149}
.
.
,
:#{149}
:..‘::..
#{149}.:.
1991
Fig. 3-30-year-old woman with insulin-dependent diabetes mellitus and neuroarthropathy had acute pain in right forefoot. A, Right lateral projection of “Tc-MDP scan shows intense uptake of radionuclide in several foot joints. B, Corresponding “1ln-WBC scan is unremarkable, suggesting no evidence of infection. The patient responded to conservative therapy and has remained well for more than 4 years, thus confirming true-negative status of lnWBC scan.
,.,
. -.
September
‘‘:
:
‘;
A
abnormalities
superimposed
such
as neuropathic
joint
dis-
between
uptake
in bone and adjacent
soft tissue,
a factor
ease, trauma, and chronic soft-tissue change, confident diagnosis made on the basis of plain radiographs or 99mTcMDp or 67Ga scans is difficult [1 , 4, 5]. Our data show that 1111nlabeled leukocytes are sensitive and specific for osteomyelitis
noted by others [8, 9, 1 2]. Another important cause of falsepositive studies in our series was the presence of advanced neuroarthropathy on the radiograph. Of 1 1 patients with such changes, uptake of ln-WBC without evidence of infection
of the
was demonstrated
foot.
preserved
Further,
in patients
the value
with
of this
imaging
neuroarthropathy,
technique
patients
is
with
soft-tissue ulcers, and those receiving antibiotics during investigation. A small minority have false-negative and falsepositive studies, which suggests that a high index of suspicion and prudent use of other noninvasive tests remain essential.
The differential uptake of ln-WBC and 99mTcMDP in a small group of patients in our series did not improve specificity further. The sensitivity
1In-WBC
scintignaphy
in our series
(79%)
was less than that found in previous investigations. Splittgerben et al. [1 3] reviewed six diabetic patients with neuroarthropathy and/or osteomyelitis and reported a sensitivity of 100%. Two larger series [8, 9] also reported no false-negative 11nWBC scans. However, the high prevalence of osteomyelitis may have contributed to these results, at least in one study [8]. Incorrect localization of 1ln-WBC uptake to soft tissues rather than bone was a factor in one false-negative study in our group. In the other two patients with false-negative 11n-
WBC scans,
necrotic
bone was identified
at surgery;
cularity and tissue necrosis may have contributed sensitivity in these cases, as has been suggested
avas-
to reduced
previously [8]. In our study, the gold standard for osteornyelitis in 14 patients was either histologic confirmation of osteomyelitis or the growth of pathogenic microorganisms from bone biopsy specimens or culture material. However, Wheat [3] has emphasized that diabetic neunoarthnopathy may be difficult to distinguish histologically from chronic osteomyelitis. Indeed, in one of our false-negative 1111n-WBC scans, resected bone was reported to show changes consistent with chronic osteomyelitis,
but no microorganisms
The relative
number
were
grown.
of false-positive 11InWBC studies in 78%) was similar to previous reports.
our series (specificity, In three of our patients, this occurred because the poor spatial resolution of the1 In-WBC scan made it difficult to distinguish
in three. By contrast,
in a recent series of
diabetic patients [8], 1ln-WBC scintignaphy cific in eight patients with neuropathic joint
group [9] also concluded
was 1 00% spedisease. Another
that sites of uninfected
neuroarthro-
pathy did not accumulate 11n-WBCs. Early (3- to 4-hn) simultaneous imaging of 99mTcMDp and 11n-WBC may contribute to false-positive studies in neuroarthropathy by virtue of “crossover” activity (mTc into the lower 11n photopeak)
[9]. However, in our series, ln-WBC scintignaphy was always performed before injection of 99mTc-MDP. A recent study [1 5] of 14 patients with Chancot joint associated with penipheral neuropathy showed that progressive neuroarthropathy of recent onset could not be distinguished from osteomyelitis on 11n-WBC scans. The authors postulated that stress frac-
tunes at such locations
that are occult on radiographs
may be
the mechanism for uptake of ln-WBCs. Consequently, we believe that accurate diagnosis of osteomyelitis in patients with neuropathic joint disease requires careful interpretation of the ln-WBC scan in combination with other imaging techniques. Serial radiographs may serve an important adjunctive role. Hyperernia and sterile inflammation have also been reported
to cause accumulation tune that
may
contributes cell-labeling erythnocytes
limit
of ln-Iabeled the
usefulness
the hypothesis eight
test.
[8], a feaHypenemia
to this phenomenon, presumably because the process involves not only leukocytes but also and platelets [21]. Accordingly, we evaluated that
only
11n-WBC
greaten than that of 99’Tc-MDP though
leukocytes of this
of nine
had osteomyelitis
patients
uptake
indicated with
(89% sensitivity),
this
proportionately
osteomyelitis. scintigraphic
Alfinding
two of four patients
with-
out osteomyelitis also had a similar pattern (50% specificity). Thus, our preliminary findings suggest that the differential
uptake
of
discriminating
11n-WBC and “Tc-MDP between
infection
may not be helpful
and nonspecific
localization
in
111INWBC
AJR:157, September1991
of leukocytes.
Downloaded from www.ajronline.org by 117.30.56.203 on 10/14/15 from IP address 117.30.56.203. Copyright ARRS. For personal use only; all rights reserved
analysis
and,
However, therefore,
these
data were
are inherently
SCINTIGRAPHY
based
limited.
on visual
Confirmation
from others who used a quantitative method may be valuable. Other factors that potentially may limit In-WBC accuracy include antibiotic therapy and coexisting soft-tissue ulcers. In our series, sensitivity and specificity were 80% and 73% (antibiotics) and 85% and 77% (ulcers), respectively. Our experience with antibiotics confirms a previous report [22] indicating that the sensitivity of ln-WBC scintigraphy was unaffected by such treatment. Osteomyelitis may complicate foot ulcers in 1 5% of diabetics [3]. Our data, based on a large subset of 35 patients, show that In-WBC scans remain valuable in this setting. Our study had several important limitations. The retrospective evaluation
of patients
referred
for diagnosis
and treatment
at a tertiary cane center contributed to selection bias. Although we are uncertain of the outcome, 25 patients excluded from analysis owing to management elsewhere were less likely to have had osteomyelitis. Nevertheless, the prevalence of osteomyelitis
in our series
was
relatively
small
(only
1 4 of 51
patients). Review of scans from patients examined during a period of 5 years introduces a potential for significant discrepancy in image quality. However, our examination protocol and gamma cameras were largely unaltered during this time. Further,
although
we encountered
minor
dissimilarities
in im-
age quality, no patients were excluded from analysis on the basis of a technically suboptimal study. In routine practice, studies are interpreted in conjunction with one another, and detailed clinical data are often available. Thus, the sensitivity and specificity of 1ln-WBC scans in our series may be an underestimation. Also, as the Tc-MDP scan was acquired after the patient had been injected with 1ln-WBCs, the 140keV photopeak of Tc may have included some crossover of activity from the lower-energy peak of ln. However, data from our laboratory indicate that, owing to a discrepancy in counting rates, this contribution is insignificant (