Journal of Orthopaedic Surgery 2011;19(2):145-50

Sensitivity, specificity and accuracy of

magnetic resonance imaging for differentiating
vertebral compression fractures caused by
malignancy, osteoporosis, and infections
ME Abdel-Wanis,1 Mohamed Tharwat Mahmoud Solyman,2 Nahla Mohamed Ali Hasan2
1
2

Department of Orthopaedic Surgery, Sohag Faculty of Medicine, Sohag University, Sohag, Egypt
Department of Diagnostic Radiology, Sohag Faculty of Medicine, Sohag University, Sohag, Egypt

ABSTRACT
Purpose. To evaluate the sensitivity, specificity
and accuracy of various magnetic resonance
imaging (MRI) features in differentiating vertebral
compression fractures caused by malignancy,
osteoporosis, and infections.
Methods. 35 men and 45 women aged 40 to 78
(mean, 59) years underwent MRI to assess the
underlying pathology of already diagnosed vertebral
compression fractures (n=152). The interval from
presentation to imaging ranged from 7 to 95 (mean,
62) days. MRI features of each vertebral compression
fracture were assessed. The sensitivity, specificity, and
accuracy for each of the MRI features were calculated.
Association between each MRI feature and various
underlying pathologies (malignancy, osteoporosis,
and infections) of vertebral compression fractures
was evaluated.
Results. Regarding these 80 patients, the MRI
diagnosis was correct in 78 and inconclusive in
2 with malignancy. MRI features suggestive of

malignant fractures were a convex posterior border

of the vertebral body, pedicle involvement, posterior
neural element involvement, an epidural mass, a
paraspinal mass, and other spinal metastases. MRI
features suggestive of osteoporotic fractures were
retropulsion, low signal intensity band, spared
normal marrow signal intensity, and the fluid
sign. MRI features suggestive of infective fractures
were contiguous vertebral involvement, end plate
disruption, disc involvement, and paraspinal and
epidural abscesses.
Conclusion. Combination of several MRI features
can provide clues to differentiate between malignant,
osteoporotic, and infective vertebral compression
fractures.
Key words: fractures, compression; infection; magnetic
resonance imaging; neoplasm metastasis; osteoporosis;
osteoporotic fractures; spinal fractures

INTRODUCTION
Vertebral

compression

fractures

are

common,

Address correspondence and reprint requests to: Dr ME Abdel-Wanis, Department of Orthopaedic Surgery, Sohag Faculty of
Medicine, Sohag, PO 82524, Egypt. E-mail: wanis307@yahoo.com

Journal of Orthopaedic Surgery

146 ME Abdel-Wanis

especially in the elderly. Differentiating malignant

from benign compression fractures based on clinical
findings, radiography, bone scans, and computed
tomography may not be sufficient, particularly for
patients without a history of trauma or malignancy.
Magnetic resonance imaging (MRI) features
for differentiating benign and malignant spinal
compression fractures have been reported.15 Most
such studies only included osteoporotic fractures as
the benign entity24; few included infective vertebral
fractures.5 Also, the relative importance, sensitivity,
specificity, and accuracy of several MRI criteria are
not fully established. We evaluated the sensitivity,
specificity and accuracy of various MRI features
in differentiating vertebral compression fractures
caused by malignancy, osteoporosis, or infections.
MATERIALS AND METHODS
From May 2003 to May 2007, 35 men and 45 women
aged 40 to 78 (mean, 59) years presented with back
or neck pain and/or a neurological deficit after no or
minor trauma (e.g. fall from standing height). They
underwent MRI to assess the underlying pathology
of the already diagnosed vertebral compression
fractures (n=152). The interval from presentation to
imaging ranged from 7 to 95 (mean, 62) days.
MRI was performed using a 1.5-tesla imager
(Gyro ACS.NT synergy coil MR device-Philips)
or a 0.2-tesla imager (Concerto Version syngo MR
2004A-Siemens).
Gadolinium-diethylenetriamine
pentacetic acid (Gd-DTPA) of 0.2 ml/kg body weight6
was administered intravenously after conventional
T1- and T2-weighted sequences are completed.
Various MRI features of each vertebral
compression fracture were assessed. They included
(1) bone marrow replacement4,7: (i) complete
replacement (no normal bone marrow signal in the
compressed vertebra), (ii) incomplete replacement
(some residual normal bone marrow signal), and
(iii) complete preservation (only normal bone
marrow signal); (2) convex posterior vertebral
border; (3) pedicle involvement; (4) posterior
element involvement; (5) other metastatic lesions
(denoted by abnormal signal intensity in the bone
marrow of other vertebrae)3; (6) retropulsion of a
posterior bone fragment (often posterosuperior
angle of the vertebral body into the spinal canal)8;
(7) a low signal intensity bandlike area on T1- and
T2-weighted images corresponding to a fracture line
or trabecular impaction; (8) a fluid sign: focal, linear
or triangular areas of high signal intensity adjacent
to the vertebral end plates on T2-weighted and short

T1 inversion recovery images1,8,9; (9) contiguous

vertebral involvement; (10) vertebral end plate
integrity; (11) vertebral disc involvement (manifest
as an abnormal signal intensity pattern: hypointense
on T1-weighted images and hyperintense on T2weighted and post-contrast enhancement images)
and reduced disc height; (12) epidural mass (anterior,
posterior or encasing); (13) a paraspinal soft-tissue
mass (anterior, posterior or both); (14) a paraspinal
abscess; and (15) an epidural abscess. Both abscesses
and soft-tissue masses exhibit low signal intensity
on T1-weighted images and high signal intensity
on T2-weighted images, but soft-tissue masses are
enhanced by Gd-DTPA whereas abscesses show ring
enhancement.2,10
The sensitivity (true positive / [true positive +
false negative] x100), specificity (true negative / [true
negative + false positive] x100), and accuracy (number
of correct MRI diagnosis / number of confirmed final
diagnosis x100) for each MRI feature were calculated.
Associations between each MRI feature and various
underlying vertebral compression fracture pathologies
were evaluated using the Chi squared test. A p value
of <0.05 was considered statistically significant.

Table 1
Diagnoses of vertebral compression fractures based on
magnetic resonance imaging (MRI)
Diagnostic
result
Correct
Inconclusive
Incorrect

No. (%) of patients

Malignant
fractures

Osteoporotic
fractures

Infective
fractures

48 (96)
2 (4)
0

16 (100)
0
0

14 (100)
0
0

Table 2
Primary tumour diagnoses of 50 patients presented with 85
malignant vertebral compression fractures
Tumour origin
Breast
Myeloma
Liver
Lung
Lymphoma
Prostate
Thyroid
Kidney
Colon
Uterus
Unknown
Total

No. (%) of
patients

No. (%) of
fractures

13 (26)
7 (14)
4 (8)
4 (8)
4 (8)
4 (8)
3 (6)
2 (4)
1(2)
1(2)
7 (14)
50 (100)

29 (34)
12 (14)
4 (5)
8 (9)
8 (9)
4 (5)
5 (6)
2 (2)
1 (1)
3 (4)
9 (11)
85 (100)

Vol. 19 No. 2, August 2011

MRI for differentiating malignant, osteoporotic, and infective vertebral compression fractures 147

Table 3
Magnetic resonance imaging (MRI) features suggestive of malignant, osteoporotic, and infective vertebral compression
fractures
MRI feature

No. (%) of fractures (n=152)

Malignant
(n=85)

Malignant fractures
Convex posterior border
Pedicle involvement
Posterior elements involvement
Epidural mass
Paraspinal mass
Other spinal metastases
Osteoporotic fractures
Retropulsion
Low signal intensity band
Spared normal marrow signal
intensity
Fluid sign
Infective fractures
Contiguous vertebral involvement
End plate disruption
Disc involvement
Paraspinal abscess
Epidural abscess

Osteoporotic
(n=34)

p Value

Infective
(n=33)

Sensitivity Specificity Accuracy

(%)
(%)
(%)

60 (71)
81 (95)
61 (72)
51 (60)
68 (80)
65 (77)

1 (3)
5 (15)
0
0
4 (12)
0

0
15 (46)
9 (27)
0
5 (15)
0

<0.0001
<0.0001
<0.0001
<0.0001
-

71
95
72
60
80
77

99
70
87
100
87
100

83
84
78
78
83
87

0
8 (9)
6 (7)

18 (53)
28 (82)
27 (79)

2 (6)
2 (6)
6 (18)

<0.0001
<0.0001
<0.0001

53
82
79

98
92
90

88
90
88

2 (2)

12 (35)

<0.0001

35

98

84

20 (24)
4 (5)
2 (2)
0
0

12 (35)
0
0
0
0

<0.0001
<0.0001
<0.0001
-

88
82
97
85
73

73
97
98
100
100

76
93
98
97
94

RESULTS
Regarding these 80 patients, the MRI diagnosis was
correct in 78 and inconclusive in 2 with malignancy
(Table 1).
In 50 patients, the underlying pathology of 85
vertebral compression fractures was malignancy
(Table 2). In 45 of them, the final diagnosis was
confirmed histopathologically through open biopsy.
In the remaining 5, it was based on progressive
deterioration of the fractured vertebra and/or new
development of spinal metastases evident with followup MRI. Indications for surgery were interactable
pain, neurological deficits, and a confirmed tissue
diagnosis.
In 16 patients, the underlying pathology of 34
vertebral compression fractures was osteoporosis. In
14 of them, the final diagnosis was based on clinical
and radiological follow-up over 6 to 12 months. In
the remaining 2, it was confirmed histopathologically.
Indication for surgery was neurological deficits.
In 14 patients, the underlying pathology of 33
vertebral compression fractures was infection. Their
final diagnoses were confirmed histopathologically.
The indications for surgery were interactable pain,
severe or progressive kyphosis, a large abscess, and
spinal instability.
MRI features suggestive of malignant fractures
were a convex posterior border of the vertebral
body, pedicle involvement, posterior neural element
involvement, an epidural mass, a paraspinal mass,
and other spinal metastases (Table 3 and Fig. 1).

29 (88)
27 (82)
32 (97)
28 (85)
24 (73)

MRI features suggestive of osteoporotic fractures

were retropulsion, a low signal intensity band,
spared normal marrow signal intensity, and the fluid
sign (Table 3 and Fig. 2). MRI features suggestive
of infective fractures were contiguous vertebral
involvement, end plate disruption, disc involvement,
paraspinal abscesses, and epidural abscesses (Table 3
and Fig. 3).
DISCUSSION
MRI is superior to other diagnostic tools in
differentiating benign from malignant compression
spinal fractures,1 based on several MRI features.29
However, only a few studies reported the sensitivity,
specificity, and accuracy of each feature,1,3 which
enable the surgeon to prioritise the MRI features in
case of contradiction. For example in our study, the
most accurate MRI feature for diagnosis of infective
fractures was disc involvement. Its sensitivity,
specificity and accuracy were 97%, 98% and 98%,
respectively. If this feature is present, the addition of
one or 2 other features suggestive of infection will be
sufficient for diagnosis.
Combination of several MRI features strongly
affirms the diagnosis of spinal compression
fractures.1 Compared to a study in 2009,1 our study
comprised a higher number of compression fractures
(152 vs. 58), and the fractures were classified into 3
categories (malignancy, osteoporosis, and infection),
rather than 2 (malignant vs. benign) categories, in

Journal of Orthopaedic Surgery

148 ME Abdel-Wanis

which the benign category (23 patients) included

trauma and infection. In our study, the final diagnosis
in 75% of patients was based on biopsy, compared to
22%.
Biopsy is the gold standard for final diagnosis
of vertebral compression fractures. A study
reported 3 patients who were provisionally
diagnosed as having osteoporotic fracture, which
was discovered by biopsy during vertebroplasty
and kyphoplasty to be malignant fractures caused
by lymphoma.11 In another study, the presumed

(a)

(b)

(c)

aetiology in 18% of cases was not confirmed

on pathological examination.12 In fact, a biopsy
picture suggestive of chronic osteomyelitis was
reported in 6 (11%) out of 66 patients diagnosed
as having osteoporotic spinal fractures.13 Because
the accuracy of percutaneous vertebral biopsy
was reported to be 80 to 90%,14 our study should
be considered accurate, as diagnoses of 75% of
patients were made by open biopsy.
Our study may help develop an MRI score
to differentiate compression fractures caused by

(d)

Figure 1 (a) Sagittal T1-weighted, (b) T2-weighted (c) short T1 inversion recovery, and (d) axial (L5) T2-weighted magnetic
resonance images (MRI) of the lumbosacral spine in a 46-year-old woman: the compression fracture of the L5 vertebral body
shows complete replacement of normal marrow signal intensity, convex posterior border (arrows), involvement of the pedicles
and transverse processes (asterisk), and metastases affecting L3 and L4 vertebrae (arrowheads). The MRI diagnosis is metastases
affecting L3, L4, and L5 vertebrae complicated with compression fracture of L5. The final diagnosis is metastatic compression
fracture of L5 originating from breast cancer.
(a)

(b)

(c)

(d)

Figure 2 Sagittal (a) T1-weighted, (b) T2-weighted, (c) short T1 inversion recovery, and (d) axial (L2 and L3) T2-weighted
magnetic resonance images (MRI) of the lumbosacral spine in a 60-year-old woman: compression fractures of L2 and L3 show
anterior wedging of the L2 vertebral body, low signal fracture line at L2 and L3, fluid sign at L3, and incomplete replacement
of normal marrow signal intensity. The MRI diagnosis is L2 and L3 acute osteoporotic compression fractures, which is also the
final diagnosis.

Vol. 19 No. 2, August 2011

(a)

(e)

MRI for differentiating malignant, osteoporotic, and infective vertebral compression fractures 149

(b)

(c)

(f)

(d)

(g)

Figure 3 Sagittal (a) T1-weighted, (b) T1-weighted after gadolinium-diethylenetriamine pentacetic acid (Gd-DTPA) injection,
(c) T2-weighted, (d) short T1 inversion recovery, and axial (e) T1-weighted, (f) T1-weighted after Gd-DTPA injection, (g) T2weighted magnetic resonance images (MRI) of the thoracic spine in a 54-year-old woman: compression fractures of T5 and T6
show contiguous vertebral involvement, complete replacement of normal bone marrow signal intensity, involvement of both
pedicles (arrows), intervening disc involvement, and an encasing epidural mass and multiple paraspinal masses (arrowhead).
The MRI diagnosis is spondylodiscitis affecting T5 and T6 and intervening disc complicated with vertebral compression fractures.
The final diagnosis is compression fractures of T5 and T6 caused by tuberculosis infection (Potts disease).

malignancy, osteoporosis, and infection. Another

study also tried to develop a scoring system based
on MRI and computed tomography to facilitate
differentiation between malignant and benign
fractures.15 The most important MRI features for
differentiation were pedicle or other posterior
element involvement, expansion to the paravertebral
region, preservation of normal bone marrow, and
continuous black line of the posterior vertebral body

margin on T2-weighted image.15 The first 2 features

suggestive of malignant fractures were given the score
of -3, whereas latter 2 features suggestive of benign
fractures were given the score +3.15 In our study, all
such features except the last had high sensitivity and
specificity. In conclusion, combination of several MRI
features can give the clue to differentiate between
malignant, osteoporotic, and infective vertebral
compression fractures.

REFERENCES
1. Pongpornsup S, Wajanawichakorn P, Danchaivijitr N. Benign versus malignant compression fracture: a diagnostic accuracy
of magnetic resonance imaging. J Med Assoc Thai 2009;92:6472.
2. Fu TS, Chen LH, Liao JC, Lai PL, Niu CC, Chen WJ. Magnetic resonance imaging characteristics of benign and malignant
vertebral fractures. Chang Gung Med J 2004;27:80815.
3. Jung HS, Jee WH, McCauley TR, Ha KY, Choi KH. Discrimination of metastatic from acute osteoporotic compression spinal
fractures with MR imaging. Radiographics 2003;23:17987.
4. Rupp RE, Ebraheim NA, Coombs RJ. Magnetic resonance imaging differentiation of compression spine fractures or vertebral
lesions caused by osteoporosis or tumor. Spine (Phila Pa 1976) 1995;20:2499504.

150 ME Abdel-Wanis

Journal of Orthopaedic Surgery

5. Tan DY, Tsou IY, Chee TS. Differentiation of malignant vertebral collapse from osteoporotic and other benign causes using
magnetic resonance imaging. Ann Acad Med Singapore 2002;31:814.
6. DeVries RM, Manne A. Cervical MRI. Part 1: a basic overview. Clin Chiropr 2003;6:13743.
7. Yuh WT, Zachar CK, Barloon TJ, Sato Y, Sickels WJ, Hawes DR. Vertebral compression fractures: distinction between benign
and malignant causes with MR imaging. Radiology 1989;172:2158.
8. Uetani M, Hashmi R, Hayashi K. Malignant and benign compression fractures: differentiation and diagnostic pitfalls on
MRI. Clin Radiol 2004;59:12431.
9. Baur A, Stabler A, Arbogast S, Duerr HR, Bartl R, Reiser M. Acute osteoporotic and neoplastic vertebral compression
fractures: fluid sign at MR imaging. Radiology 2002;225:7305.
10. Vaccaro AR, Shah SH, Schweitzer ME, Rosenfeld JF, Cotler JM. MRI description of vertebral osteomyelitis, neoplasm, and
compression fracture. Orthopedics 1999;22:6773.
11. Shindle MK, Tyler W, Edobor-Osula F, Gardner MJ, Shindle L, Toro J, et al. Unsuspected lymphoma diagnosed with use of
biopsy during kyphoplasty. J Bone Joint Surg Am 2006;88:27214.
12. Schoenfeld AJ, Dinicola NJ, Ehrler DM, Koerber A, Paxos M, Shorten SD, et al. Retrospective review of biopsy results
following percutaneous fixation of vertebral compression fractures. Injury 2008;39:32733.
13. Allen RT, Kum JB, Weidner N, Hulst JB, Garfin SR. Biopsy of osteoporotic vertebral compression fractures during
kyphoplasty: unsuspected histologic findings of chronic osteitis without clinical evidence of osteomyelitis. Spine (Phila Pa
1976) 2009;34:148691.
14. Nourbakhsh A, Grady JJ, Garges KJ. Percutaneous spine biopsy: a meta-analysis. J Bone Joint Surg Am 2008;90:17225.
15. Yuzawa Y, Ebara S, Kamimura M, Tateiwa Y, Kinoshita T, Itoh H, et al. Magnetic resonance and computed tomography-based
scoring system for the differential diagnosis of vertebral fractures caused by osteoporosis and malignant tumors. J Orthop
Sci 2005;10:34552.