AJR 2004; 182:1405-1410
© American Roentgen Ray Society
Discrimination of Tuberculous Spondylitis from Pyogenic Spondylitis on MRI
Na-Young Jung1,
Won-Hee Jee1,
Kee-Yong Ha2,
Chun-Kun Park3 and
Jae-Young Byun1
1 Department of Diagnostic Radiology, Kangnam St. Mary's Hospital, The Catholic
University of Korea, 505 Banpodong, Seocho-ku, Seoul 137-040, Korea.
2 Department of Orthopedic Surgery, Kangnam St. Mary's Hospital, The Catholic
University of Korea, Seoul, Korea.
3 Department of Neurosurgery, Kangnam St. Mary's Hospital, The Catholic
University of Korea. Seoul, Korea.
Received October 29, 2003;
accepted after revision November 20, 2003.
Address correspondence to W.-H. Jee
(whjee{at}catholic.ac.kr).
Abstract
OBJECTIVE. The purpose of this study was to determine the accuracy
of MRI for discrimination between tuberculous spondylitis and pyogenic
spondylitis.
MATERIALS AND METHODS. MR images of 52 patients who had MRI of the
spine and confirmed spondylitis were retrospectively reviewed. After review of
medical records, we compared MRI findings in 20 patients with tuberculous
spondylitis and 20 patients with pyogenic spondylitis. Statistical analysis
was performed with the chi-square test.
RESULTS. The reviewer identified tuberculous spondylitis with
sensitivity, specificity, and accuracy of 100% (20/20), 80% (16/20), and 90%
(36/40), and pyogenic spondylitis with sensitivity, specificity, and accuracy
of 80% (16/20), 100% (20/20), and 90% (36/40), respectively. The patients with
tuberculous spondylitis had a significantly higher incidence of MRI findings
as follows (p < 0.05): a well-defined paraspinal abnormal signal
(95% [19/20] in tuberculous vs 25% [5/20] in pyogenic), a thin and smooth
abscess wall (95% [19/20] vs 15% [3/20]), combination of both findings (90%
[18/20] vs 0% [0/20]), presence of paraspinal or intraosseous abscess (95%
[19/20] vs 50% [10/20]), subligamentous spread to three or more vertebral
levels (85% [17/20] vs 40% [8/20]), involvement of multiple vertebral bodies
(60% [12/20] vs 25% [5/20]), thoracic spine involvement (40% [8/20] vs 10%
[2/20]), and hyperintense signal on T2-weighted images (95% [19/20] vs 65%
[13/20]).
CONCLUSION. MRI was accurate for differentiation of tuberculous
spondylitis from pyogenic spondylitis.
Introduction
Infectious spondylitis is defined as an infection by a specific organism of
one or more components of the spine, namely the vertebra, intervertebral
discs, paraspinal soft tissues, and epidural space
[1]. It is important to
differentiate tuberculous spondylitis from pyogenic spondylitis because proper
treatment of the different types can reduce the rate of disability and
functional impairment [2,
3]. However, it is sometimes
difficult to differentiate these two types clinically and radiographically
[1,
2].
The purpose of this study was to determine the accuracy of MRI in
discriminating between tuberculous spondylitis and pyogenic spondylitis.
Materials and Methods
MR images of the spines of 52 patients with infectious spondylitis at our
institution over an 8-year period were retrospectively reviewed by a
musculoskeletal radiologist with 9 years' experience without knowledge of
clinical history or pathology results. After chart review by one author, 12
patients with old tuberculous infections were excluded because of the absence
of evidence of active disease such as abnormal signal intensity or
enhancement. All 20 cases of tuberculous spondylitis (six men, 14 women) and
20 cases of pyogenic spondylitis (13 men, seven women) were confirmed by
biopsy. The mean ages of patients with tuberculous spondylitis and pyogenic
spondylitis were 41 years (range, 17–77 years) and 58 years (range,
25–80 years), respectively. The mean interval from presentation to MRI
was 35 weeks (range, 2 weeks–36 months) in patients with tuberculous
spondylitis and 16 weeks (range, 5 days–24 months) in patients with
pyogenic spondylitis.
MRI was performed on 1.5-T imagers (Signa, General Electric Medical
Systems) in 19 patients and (Magnetom Vision, Siemens) in 21 patients, using a
surface coil or spine coil. Axial and sagittal T1-weighted MR images (TR
range/TE range, 350–650/11–30) and fast spin-echo or turbo
spin-echo T2-weighted images (3,000–4,000/76–108) were obtained.
In addition, axial and sagittal fat-suppressed T1-weighted images
(350–800/11–30) were obtained after IV infusion of 0.1 mmol/kg of
gadopentetate dimeglumine. Typical MR parameters were as follows: field of
view, 15–20 cm for axial plane and 30–35 cm for sagittal plane;
number of excitations, 2; matrix size, 256 x 192; slice thickness, 4 mm;
intersection gap, 1 mm; and echo-train length, 8–16.
The reviewer evaluated the presence or absence of individual imaging
criteria and made an overall assessment of the type of spondylitis. The margin
of paraspinal abnormal signal, the appearance of the abscess walls, the extent
of subligamentous spread, horizontal bandlike sparing of the body, involvement
of multiple vertebral bodies, the involvement of the thoracic spine, entire
body involvement, and the signal intensity of involved vertebral bodies were
evaluated. The abscess wall was assessed by the reviewer on the basis of the
contrast-enhanced images. The signal intensity in the marrow of abnormal
vertebrae was considered hypointense, isointense, or hyperintense by
comparison with the signal intensity of normal vertebrae in the same patient
on T1- and T2-weighted images. Statistical analysis was performed with the
chi-square test.
Results
The distinction between tuberculous spondylitis (Fig.
1A,
1B,
1C,
1D) and pyogenic spondylitis
(Fig. 2A,
2B,
2C,
2D) could be made on the basis
of MRI findings (Table 1). The
reviewer identified tuberculous spondylitis with sensitivity, specificity, and
accuracy values of 100%, 80%, and 90%, respectively. For pyogenic spondylitis,
the corresponding values were 80%, 100%, and 90%.
View larger version (136K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1A. —Typical tuberculous spondylitis in 53-year-old woman.
Sagittal T1-weighted image (TR/TE, 547/12) shows heterogeneously hypointense
signal (arrows) in T8–T9 vertebral bodies with epidural mass
and subligamentous spread (arrowheads) from T7 to T10.
|
|
View larger version (121K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1B. —Typical tuberculous spondylitis in 53-year-old woman. T8 and
T9 vertebral bodies are heterogeneously hyperintense (arrows) on
sagittal turbo spin-echo T2-weighted image (3,000/112; echo-train length,
15).
|
|
View larger version (128K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1C. —Typical tuberculous spondylitis in 53-year-old woman.
Sagittal fat-suppressed contrast-enhanced T1-weighted image (627/12) shows
heterogeneous enhancement (arrows) of T8–T9 vertebral bodies.
Intraosseous abscess (asterisk) is present.
|
|
View larger version (163K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1D. —Typical tuberculous spondylitis in 53-year-old woman. Axial
fat-suppressed contrast-enhanced T1-weighted image (740/15) shows well-defined
paraspinal abnormal enhancement (arrows) and thin and smooth rim
enhancement of paraspinal abscess (asterisk). Prominent meningeal
enhancement is present.
|
|
View larger version (118K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B. —Typical pyogenic spondylitis in 53-year-old man. On sagittal
turbo spin-echo T2-weighted image (3,200/99; echo-train length, 11) L4 and L5
vertebral bodies are isointense (arrows) to adjacent normal
vertebrae.
|
|
View larger version (114K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2C. —Typical pyogenic spondylitis in 53-year-old man. Sagittal
fat-suppressed contrast-enhanced T1-weighted image (400/12) shows diffuse
heterogeneous enhancement (arrows) in the L4 and L5 vertebral bodies.
Abscess (asterisk) is present in L4–5 disk space extending to
L5 vertebral body.
|
|
View larger version (166K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2D. —Typical pyogenic spondylitis in 53-year-old man. Axial
fat-suppressed contrast-enhanced T1-weighted image (800/15) shows thick and
irregular rim enhancement of paraspinal abscess (asterisk).
Ill-defined paraspinal abnormal enhancement (arrows) is present.
|
|
The incidence of the following MRI findings was significantly higher in
patients with tuberculous spondylitis than in those with pyogenic spondylitis
(p < 0.05): a well-defined paraspinal abnormal signal (95% in
tuberculous vs 25% in pyogenic) (Fig.
3A,
3B,
3C,
3D), a thin and smooth abscess
wall (95% vs 15%), combination of both findings (90% vs 0%), presence of
paraspinal or intraosseous abscess (95% vs 50%), subligamentous spread or more
three vertebral levels (85% vs 40%), involvement of multiple vertebral bodies
(60% vs 25%), and thoracic spine involvement (40% vs 10%).
View larger version (113K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3B. —Pyogenic spondylitis in 66-year-old woman. On sagittal fast
spin-echo T2-weighted image (3,000/112; echo-train length, 8) L2 and L3
vertebral bodies are isointense (arrows) to adjacent normal vertebrae
with well-defined high signal (asterisk).
|
|
View larger version (121K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3C. —Pyogenic spondylitis in 66-year-old woman. Sagittal
fat-suppressed contrast-enhanced T1-weighted image (600/11) shows
heterogeneous enhancement (arrows) of L2 and L3 vertebral bodies with
thin- and smooth-walled abscess (asterisk).
|
|
The incidence of the following MRI findings was significantly higher in
patients with pyogenic spondylitis than in those with tuberculous spondylitis
(p < 0.05): an ill-defined paraspinal abnormal signal (70% in
pyogenic vs 5% in tuberculous) (Fig.
4A,
4B,
4C,
4D), absence of paraspinal or
intraosseous abscess (50% vs 5%), subligamentous spread to fewer than three
vertebral levels or without subligamentous spread (60% vs 15%), a thick and
irregular abscess wall (35% vs 0%), a horizontal bandlike sparing of the body
(25% vs 0%), and involvement of two or fewer vertebral bodies (75% vs
40%).
View larger version (138K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4A. —Tuberculous spondylitis in 69-year-old woman. Sagittal
T1-weighted image (TR/TE, 650/13) shows hypointense T12 vertebral body with
compression. Adjacent hypointense paraspinal abnormal signal (black
arrows) and epidural mass (white arrows) are present.
|
|
View larger version (145K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4B. —Tuberculous spondylitis in 69-year-old woman. T12 vertebral
body is hypointense with internal high signal (arrows) on sagittal
fast spin-echo T2-weighted image (2,500/118; echo-train length, 14).
|
|
View larger version (123K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4C. —Tuberculous spondylitis in 69-year-old woman. Sagittal
fat-suppressed contrast-enhanced T1-weighted image (650/13) shows
heterogeneous enhancement of T12 vertebral body with paraspinal abnormal
enhancement (arrows) anteriorly. Epidural abscess (asterisk)
is present.
|
|
All patients with spondylitis showed a hypointense to isointense signal on
T1-weighted images. No significant difference in heterogeneous signal was seen
on T1-weighted images (100% in tuberculous vs 85% in pyogenic, p =
0.115) or high signal foci on T1-weighted images (20% vs 10%, p =
0.331). Hyperintense signal on T2-weighted images was more commonly observed
in tuberculous spondylitis (95% in tuberculous vs 65% in pyogenic, p
< 0.05), whereas isointense signal on T2-weighted images was more commonly
observed in pyogenic spondylitis (35% in pyogenic vs 0% in tuberculous,
p = 0.004). No significant difference in hypointense signal was seen
on T2-weighted images (5% in tuberculous vs 0% in pyogenic, p = 0.
500) or heterogenous signal on T2-weighted images (100% vs 95%, p =
0.500). Contrast enhancement pattern showed no significant difference in
either the heterogeneous pattern (100 in tuberculous vs 80% in pyogenic,
p = 0.053) or the homogeneous pattern (0% vs 20%, p =
0.053).
In tuberculous spondylitis, thoracic or thoracolumbar involvement was
observed in 40% of cases and lumbar involvement in 50% of cases, whereas in
pyogenic spondylitis, thoracic involvement was observed in 10% of cases and
lumbar involvement was observed in 70% of cases. The involvement of the
posterior element (90% in tuberculous vs 70% in pyogenic, p = 0.118)
and epidural extension (100% vs 90%, p = 0.244) were not
significantly different in tuberculous and pyogenic spondylitis. No difference
was present in the involvement of intervertebral disk (80% in tuberculous vs
90% in pyogenic, p = 0.171) and disk space narrowing (55% in
tuberculous vs 45% in pyogenic, p = 0.376).
The longest contiguous involvement of the spine in tuberculous spondylitis
was associated with five vertebral bodies. One patient with tuberculous
spondylitis had a skip lesion involving the thoracic and lumbar spine
independently, and in one patient the tuberculous spondylitis involved only
one vertebral body.
Discussion
The incidence of typical acute vertebral osteomyelitis has decreased
because of the more wide-spread use of antibiotics. However, tuberculous
spondylitis is still a frequent cause of infectious spondylitis in endemic
regions and is increasing in prevalence because of the resurgence of
tuberculosis during the past decade, especially in patients who are
immunocompromised
[1–4].
Tuberculosis of the spine accounts for more than 50% of musculoskeletal
tuberculosis. The importance of early diagnosis and prompt treatment of
infectious spondylitis based on a specific diagnosis cannot be overemphasized
in minimizing the residual spinal deformity or permanent neurologic deficit
[2–5].
Differentiation between tuberculous and pyogenic spondylitis is difficult
clinically and radiographically. MRI has been reported to be useful in the
early detection of spondylitis
[2,
6–8].
To our knowledge, there have been a few reports
[2,
9,
10] on differential MRI
findings between tuberculous spondylitis and pyogenic spondylitis.
Rim enhancement of abscess on MRI is reportedly suggestive of tuberculous
spondylitis [1,
2,
4,
7,
10]. However, rim enhancement
was observed in both tuberculous and pyogenic spondylitis in this study. The
two most reliable MRI findings suggesting tuberculous spondylitis in our study
were thin and smooth enhancement of the abscess wall and well-defined
paraspinal abnormal signal, whereas thick and irregular enhancement of abscess
wall and ill-defined paraspinal abnormal signal were suggestive of pyogenic
spondylitis. Thus, contrast-enhanced MRI was essential in the differentiation
of these two types of spondylitis. Hong et al.
[11] reported that in
tuberculous arthritis the margins of extraarticular lesions were smoother and
the abscess walls were thinner and smoother than in pyogenic arthritis.
Regarding the margin of the soft-tissue abnormal signal and the appearance of
the abscess wall, tuberculous and pyogenic spondylitis presented MRI findings
similar to those of tuberculous and pyogenic arthritis, respectively. We
assumed that the relative late phase and chronic course of tuberculous
spondylitis contributed to the smoother margin of the paraspinal abnormal
signal and a thinner and smoother abscess wall. The minimal inflammation of
tuberculous abscess may also contribute to the thin and smooth appearance of
abscess wall.
It was recently reported that signal intensity was of limited value in
differentiating tuberculous arthritis from pyogenic arthritis
[11]. In contrast, in this
study hyperintense signal on T2-weighted images were more common in
tuberculous spondylitis than in pyogenic spondylitis. This discrepancy between
the studies could be due to the use of various MRI units in the previous
study, whereas only 1.5-T scanners at one institution were used in this
study.
A lack of proteolytic enzymes in the Mycobacterium as compared
with pyogenic infection has been proposed as the cause of relative
preservation of the intervertebral disks
[2,
9,
12]. However, in this study,
disk space narrowing was observed in 55% (11/20) of the patients with
tuberculous spondylitis, which was similar to that for 45% (9/20) of the
patients with pyogenic spondylitis. We assumed that a longer interval existed
from presentation to MRI in tuberculous spondylitis than pyogenic spondylitis
in our study. Disk space narrowing was prominent in 72% (21/29) of patients
with tuberculous spondylitis in the study by Liu et al.
[13].
Subligamentous spread to three or more vertebral levels was frequent in
tuberculous spondylitis in our study, which is consistent with previous
reports [8,
10]. The involvement of the
thoracic spine and multiple vertebral bodies was significantly higher in our
patients with tuberculous spondylitis than in patients with pyogenic
spondylitis, which is also consistent with previous reports
[5,
14]. Horizontal bandlike
sparing of the body was exclusively observed in 25% of cases of pyogenic
spondylitis, although it was observed in a minority of the patients with
pyogenic spondylitis. In contrast to previous reports
[5,
7], lumbar involvement was as
common as thoracic and thoracolumbar involvement in our patients with
tuberculous spondylitis. We assume that lumbar involvement of tuberculous
spondylitis probably is not as uncommon as considered previously. Involvement
of the posterior element has been reported in tuberculous spondylitis and very
uncommonly in pyogenic spondylitis
[2,
4,
9]. According to a recent
report [15], posterior element
tuberculosis is not as rare (at 24%) as previously reported. In our study,
abnormal signal of the posterior element was observed in 90% of patients with
tuberculous spondylitis and in 70% of patients with pyogenic spondylitis, but
this difference was not significant. We consider that more common abnormal
signals of the posterior element in this study could be related to the use of
fat-suppressed contrast-enhanced T1-weighted images with 1.5-T MR scanners.
Skip lesions have been reported as suggestive MRI findings of tuberculous
spondylitis [1,
2,
7,
14,
15], and such a lesion was
observed in one patient with tuberculous spondylitis in this study.
Several limitations are associated with the study reported here. MR images
were interpreted by only one experienced musculoskeletal radiologist. Thus,
interobserver variability and accuracy associated with less experienced
radiologists was not assessed. The sample sizes were small. The reviewer knew
that all cases of spondylitis were either tuberculous or pyogenic, which may
have increased the sensitivity of the MRI diagnosis for both conditions. The
time-of-presentation factor may influence the margin of paraspinal abnormal
signal and the appearance of abscess wall; therefore, a better study would be
to compare same time to clinical presentation cases.
In conclusion, MRI was accurate for differentiation of tuberculous
spondylitis from pyogenic spondylitis. A well-defined paraspinal abnormal
signal, a thin and smooth abscess wall, subligamentous spread to three or more
vertebral levels, and multiple vertebral or entire body involvement were more
suggestive of tuberculous spondylitis than pyogenic spondylitis.
Acknowledgments
We thank Thomas R. McCauley for his review of the manuscript.
References
- Sharif HS, Morgan JL, al Shahed MS, al Thagafi MY. Role of CT and
MR imaging in the management of tuberculous spondylitis. Radiol
Clin North Am 1995;33:787
–804[Medline]
- Arizono T, Oga M, Shiota E, Honda K, Sugioka Y. Differentiation of
vertebral osteomyelitis and tuberculous spondylitis by magnetic resonance
imaging. Int Orthop1995; 19:319
–322[Medline]
- Moore SL, Rafii M. Imaging of musculoskeletal and spinal
tuberculosis. Radiol Clin North Am2001; 39:329
–342[Medline]
- Shanley DJ. Tuberculosis of the spine: imaging features.
AJR 1995;164:659
–664[Abstract/Free Full Text]
- Alothman A, Memish ZA, Awada A, et al. Tuberculous spondylitis:
analysis of 69 cases from Saudi Arabia. Spine2001; 26:E565
–570[Medline]
- Stabler A, Reiser MF. Imaging of spinal infection.
Radiol Clin North Am2001; 39:115
–135[Medline]
- Akman S, Sirvanci M, Talu U, Gogus A, Hamzaoglu A. Magnetic
resonance imaging of tuberculous spondylitis.
Orthopedics2003; 26:69
–73[Medline]
- Griffith JF, Kumta SM, Leung PC, Cheng JC, Chow LT, Metreweli C.
Imaging of musculoskeletal tuberculosis: a new look at an old disease.
Clin Orthop2002; 398:32
–39
- Smith AS, Weinstein MA, Mizushima A, et al. MR imaging
characteristics of tuberculous spondylitis vs vertebral osteomyelitis.
AJR 1989; 153:399
–405[Abstract/Free Full Text]
- Thrush A, Enzmann D. MR imaging of infectious spondylitis.
AJNR 1990;11:1171
–1180[Abstract]
- Hong SH, Kim SM, Ahn JM, Chung HW, Shin MJ, Kang HS. Tuberculous
versus pyogenic arthritis: MR imaging evaluation.
Radiology 2001;218
: 848–853[Abstract/Free Full Text]
- Gouliamos AD, Kehagias DT, Lahanis S, et al. MR imaging of
tuberculous vertebral osteomyelitis: pictorial review. Eur
Radiol 2001;11:575
–579[Medline]
- Liu GC, Chou MS, Tsai TC, Lin SY, Shen YS. MR evaluation of
tuberculous spondylitis. Acta Radiol1993; 34:554
–558[Medline]
- Moorthy S, Prabhu NK. Spectrum of MR imaging findings in spinal
tuberculosis. AJR2002; 179:979
–983[Free Full Text]
- Narlawar RS, Shah JR, Pimple MK, Patkar DP, Patankar T, Castillo M.
Isolated tuberculosis of posterior elements of spine: magnetic resonance
imaging findings in 33 patients. Spine2002; 27:275
–281[Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
|
|
 
A. Pourbagher, M. A. Pourbagher, L. Savas, T. Turunc, Y. Z. Demiroglu, I. Erol, and D. Yalcintas
Epidemiologic, Clinical, and Imaging Findings in Brucellosis Patients with Osteoarticular Involvement
Am. J. Roentgenol.,
October 1, 2006;
187(4):
873 - 880.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
