Evaluation of the diagnostic utility of spinal magnetic resonance imaging in axial spondylarthritis.

ARTHRITIS & RHEUMATISM
Vol. 60, No. 5, May 2009, pp 1331–1341
DOI 10.1002/art.24493
© 2009, American College of Rheumatology
Evaluation of the Diagnostic Utility of Spinal
Magnetic Resonance Imaging in Axial Spondylarthritis
A. N. Bennett,1 A. Rehman,2 E. M. A. Hensor,3 H. Marzo-Ortega,3 P. Emery,3
and D. McGonagle4
were of low diagnostic utility for SpA, but >3 RLs
(likelihood ratio [LR] 12.4) and severe RLs (LR infinite) in younger subjects were highly diagnostic of SpA.
Posterior element BME lesions of mild or moderate
grade were also highly diagnostic of SpA (LR 14.5). The
most common diagnostic confusion was between SpA
and DA, since both had RLs present and the presence/
absence of degenerative discs did not change the diagnostic assessment.
Conclusion. This study confirms the high diagnostic utility of MRI in axial SpA, with severe or
multiple RLs evident on MRI being characteristic in
younger patients and mild/moderate posterior element
lesions being specific for SpA. However, MRI lesions
previously considered to be characteristic of SpA could
also be found frequently in patients with DA and
patients with malignancy, and therefore such lesions
should be interpreted with caution, particularly in older
patients.
Objective. Magnetic resonance imaging (MRI) is
increasingly used for the diagnosis of axial spondylarthritis (SpA), but it is unknown whether characteristic
lesions are actually specific for SpA. This study was
undertaken to compare MRI patterns of disease in
active SpA, degenerative arthritis (DA), and malignancy.
Methods. Fat-suppressed MRI of the axial skeleton was performed on 174 patients with back pain and
11 control subjects. Lesions detected by MRI, including
Romanus lesions (RLs) and end-plate, diffuse vertebral
body, posterior element, and spinous process bone
marrow edema (BME) lesions, were scored in a blinded
manner. An imaging diagnosis was given based on MRI
findings alone, and this was compared with the goldstandard treating physician’s diagnosis.
Results.The physician diagnosis was SpA in 64
subjects, DA in 45 subjects, malignancy in 45 subjects,
other diagnoses in 20 subjects, and normal in 11 subjects. There was 72% agreement between the imaging
diagnosis and physician diagnosis. End-plate edema,
degenerative discs, and RLs were frequently observed in
patients with any of the 3 major diagnoses. Single RLs
In recent years, magnetic resonance imaging
(MRI) has become established as a diagnostic tool for
spondylarthritis (SpA), particularly in early sacroiliac
(SI) joint disease. This is principally because the diagnostic radiographic changes in the SI joints and spine
usually develop slowly (1). Fat-suppressed MRI enables
identification of osteitis at disease presentation and has
the potential to transform the diagnosis of axial SpA,
since much of the axial pathology is related to a diffuse
perifibrocartilage osteitis, either in the SI joints or at the
entheses in the spine (2–4).
Many features of SpA that have been identified
on MRI, including Romanus lesions (RLs), end-plate
lesions, diffuse vertebral body lesions, posterior element
(facet and pedicle) lesions, and spinous process bone
marrow edema (BME) lesions, are recognized as characteristic of SpA (5–7). It is generally assumed that these
MRI patterns of disease can be used to distinguish SpA
Dr. Hensor’s work is supported in part by an ARC program
grant.
1
A. N. Bennett, MBChB, MRCP, MSc, MFSEM: Leeds
Institute of Molecular Medicine and Chapel Allerton Hospital, Leeds,
and Defence Medical Rehabilitation Centre–Headley Court, Epsom,
UK; 2A. Rehman, MBBS, MRCP, FRCR: American Hospital Dubai,
Oud Metha, Dubai, United Arab Emirates; 3E. M. A. Hensor, PhD, H.
Marzo-Ortega, MRCP, PhD, P. Emery, MA, MD, FRCP: Leeds
Institute of Molecular Medicine and Chapel Allerton Hospital, Leeds,
UK; 4D. McGonagle, PhD, FRCPI: Leeds Institute of Molecular
Medicine and Chapel Allerton Hospital, Leeds, UK, and Merlin Park
Hospital, Galway, Ireland.
Address correspondence and reprint requests to D. McGonagle, PhD, FRCPI, Academic Section of Musculoskeletal Disease,
Leeds Institute of Molecular Medicine, Chapel Allerton Hospital,
Chapeltown Road, Leeds LS7 4SA, UK. E-mail: d.g.mcgonagle@
leeds.ac.uk.
Submitted for publication August 21, 2008; accepted in
revised form February 9, 2009.
1331
1332
BENNETT ET AL
from other diseases. Indeed, these lesions are being used
to diagnose axial SpA early in its onset (3,8) and might
influence early initiation of biologic therapy; for this
purpose, several systems of scoring spinal BME lesions
have been established (6,9–12). However, the specificity
of these spinal BME lesions for SpA and their diagnostic
utility for SpA have not been confirmed. Despite this,
axial imaging continues to assume increasing importance
in SpA. Furthermore, acute inflammation in axial SpA
affects only the spine and spares the SI joints in 23% of
cases (13), so there is a need to better define the role of
MRI in this group.
However, distinguishing spinal MRI patterns of
different diseases may not be simple. We have previously
studied distal interphalangeal (DIP) joint disease in 2
distinctly different conditions, psoriatic arthritis (PsA)
and osteoarthritis (OA), and found that MRI patterns of
disease were difficult to differentiate, because the entheseal and bone-related changes were similar (14). We
therefore assessed the pattern of MRI lesions typically
noted in SpA across 3 major causes of chronic back pain,
the clinically confirmed European Spondylarthropathy
Study Group (ESSG) criteria for SpA (15), degenerative
spinal arthritis (DA), and spinal malignancy/metastases,
in comparison with a normal control group of subjects
without back pain. A wide age group was studied to
capture all ages at presentation of SpA (16–18), and
particularly to capture the 30–50-year-old age group, in
whom limited data suggest that early degenerative
changes on MRI can be mistaken for axial SpA (19).
The diagnostic utility of these characteristic
changes in SpA was analyzed. In particular, MRI patterns of disease were compared between conditions to
ascertain whether distinct anatomic patterns of disease
exist, and whether, as highlighted in the research agenda
of the Outcome Measures in Rheumatology Clinical
Trials (OMERACT) 7 (20), BME is specific for the
inflammation in SpA.
PATIENTS AND METHODS
Cohort. Patients were recruited from Calderdale Royal
Hospital (CRH), West Yorkshire, UK. Ethics committee approval was obtained, and all living patients gave their written
informed consent. Potential patients were identified by clinical
radiology staff, via the Patient Archiving and Communication
Systems at CRH, as having had MRI of the whole spine with or
without imaging of the SI joints for back pain in the last 5
years. Patients who met these criteria were invited to participate in the study. A control group of subjects with no back pain
was also recruited.
The diagnosis that was given by the treating physician
was considered the gold standard. The gold-standard physician
diagnosis was based on a history review, physical examination,
nonradiologic and radiologic investigations, histologic findings
(when available), and clinical outcome data. Patients’ case
records, with the benefit of up to 5 years of outcome data, were
then evaluated to establish whether the treating physician’s
diagnosis was correct.
In the SpA cohort (n ⫽ 64), all patients met the ESSG
criteria for SpA (15) prior to MRI scanning. Specific diagnoses
in this group were ankylosing spondylitis (AS) in 20 patients,
undifferentiated SpA in 17 patients, inflammatory bowel disease–
related SpA in 14 patients, PsA in 9 patients, and reactive
arthritis in 4 patients, with a mean disease duration of 8.5 years
(AS patients, mean disease duration 15.2 years). Ninety-four
percent of the patients with AS and 68% of all SpA patients
were HLA–B27 positive. All SpA patients had clinically active
disease, and MRI scans had been performed to assess ongoing
disease activity. Three patients were receiving anti–tumor
necrosis factor therapy at the time of the scan but still had
clinically active disease.
Sample size. Power calculations were performed
using specificity and sensitivity nomograms (21). The prevalence of SpA in the present study cohort was 35%. A sample
size of 150 allowed an accurate evaluation of MRI lesions in
the diagnosis of SpA, with measurement of specificity and
sensitivity at the 95% level with 6% precision at 95% confidence levels.
MRI scoring. A total of 185 patients had MR images
of the whole spine, comprising sagittal T1 and STIR (fatsuppressed) sequences. In addition, MR images of the SI
joints, comprising coronal/oblique T1 and STIR sequences,
were obtained in 89 patients (48%), of whom 91% had SpA
and 26% had other diagnoses. This reflects standard clinical
practice, since many patients with DA or malignant spinal
disease do not have the SI joints scanned.
Whole-spine MRI scan protocols were standardized
across the different patient groups. All MR scans were scored
by consensus by 2 experienced observers (AR and DM), who
were blinded to the patients’ details and clinical characteristics.
The Leeds Scoring System (6,12) was used to score spinal
BME as well as BME of the SI joints on STIR sequences. BME
was defined, according to the OMERACT definitions (22,23),
as low signal intensity on the T1 sequence and high/
intermediate bone marrow signals with irregular contour on
the STIR image, when compared with normal bone marrow
signals coming from the sacral interforaminal area in the SI
joint scans or the center of an unaffected vertebra in the spinal
scans, similar to the Spondyloarthritis Research Consortium of
Canada definition (9).
In the SI joints, a semiquantitative global scale was
used to score the severity of BME according to the extent of
the quadrant affected, as follows: grade 0 ⫽ normal, grade 1 ⫽
⬍25%, grade 2 ⫽ 25–75%, or grade 3 ⫽ ⬎75%. This system
has been used in previous studies (6,12) and, as reported in
OMERACT 7 (20,24), is considered equivalent to alternative
scoring systems. In addition to the scoring of active BME
lesions, structural changes, as shown by the presence or
absence of erosions and ankylosis, were also recorded, with
erosions identified as subchondral defects with a low signal on
T1-weighted images, and ankylosis defined as complete bridging of the SI joints with total loss of joint space on T1
sequences.
SPINAL MRI DIAGNOSIS OF AXIAL SpA
For the spine, the Leeds Scoring System was used to
identify different patterns of vertebral lesions, since some
patterns of acute vertebral inflammation are thought to also
occur frequently in DA (9,19). The scoring system was modified to grade the severity of the lesions, since it was considered
important to use a semiquantitative system rather than a
dichotomous system to grade the severity of spinal lesions,
because severity of BME lesions in the SI joints has been
shown to play a role in the early diagnosis of AS (3). We
decided to modify the Leeds Scoring System rather than use
other available scoring systems (9–11), since other systems do
not include posterior elements (25), which is an area that
seems to be highly relevant to SpA, as has also been noted by
other authors (26,27).
A training session was arranged prior to these assessments, in which multiple scans were reviewed to recognize
different examples of the wide range of spinal pathologic
findings described in the literature for SpA, malignancy, and
DA (5–7,28–33). Readers discussed the scans and agreed on
definitions. All lesions were identified as those displaying a
high signal on STIR sequences, as described above. Specific
pathologic findings were identified for each type of lesion. RLs
were defined as BME lesions arising from any corner of any
vertebrae from the inferior C2 to superior S1, extending
toward the center of the vertebrae but not occupying more
than 50% of the end plate. End-plate lesions were those arising
from either the superior or inferior end plates from the inferior
C2 to superior S1, extending across more than 50% of the end
plate beyond the center of the vertebrae. Posterior element
lesions were those arising from either the pedicle or facet joint
from C2 to L5, with extension toward either the vertebral body
or the spinous process. Spinous process lesions were those
arising from the spinous process of any vertebrae from C2 to
L5, with possible extension toward the facet joint/pedicle or
toward the intraspinous ligaments. Diffuse vertebral body
edema lesions were identified as those showing a relatively
homogeneous high signal within the whole of the vertebral
body from any vertebrae from C2 to S1.
All spinal lesions were graded using a semiquantitative scale, ranging from 0 to 3 (0 ⫽ normal, 1 ⫽ mild, 2 ⫽
moderate, 3 ⫽ severe) according to the intensity and extent of
the BME signal affecting the anatomic structure involved, i.e.,
quadrants of vertebrae for RLs, half vertebrae for end-plate
lesions, whole vertebrae for diffuse vertebral edema lesions,
facet and pedicle for posterior element lesions, and spinous
process for spinous process lesions. This grading system (range
0–3 for mild–severe) is very similar to that used in the Berlin
Scoring System (34) and the ASpireMRI BME scoring system
(10), the difference being that the modifications in the Leeds
Scoring System allowed additional scoring of posterior elements, including spinous processes, and allowed differentiation
of vertebral lesions, including RLs, end-plate lesions, or diffuse
vertebral edema lesions. Other frequently reported lesions
were those typical of malignancy, i.e., diffuse patchy uptake
with low signal on T1 and medium/high signal on STIR, bone
destruction, expansile/lytic/sclerotic lesions, or soft-tissue
masses (30). Typical degenerative changes in the degenerative
discs or end-plate lesions were also recorded (31,35).
In those cases where both the spine images and SI joint
images were available, they were scored separately, reflecting
the fact that in the investigation of back pain across different
1333
specialties in clinical practice, it is common for spinal MRI to
be performed without SI joint MRI sequences. Diagnoses
based on spinal images were given separately from those based
on SI joint scans, followed by an overall diagnosis that was
determined on the basis of the 2 combined, when SI joint scans
were also available. The overall imaging diagnosis was compared with the gold-standard treating physician’s diagnosis,
and the frequency and pattern of lesions on MRI were
compared with the gold-standard physician diagnosis to assess
diagnostic utility.
Reliability of scoring. The reliability of the MRI
scoring was assessed for both the SI joints and the spine
separately, on 25 scans that were read at 2 time points, 1 week
apart. Agreement on overall diagnosis was 100% for readings
of the SI joint scans and 100% for readings of the spine scans
(overall quadratic-weighted kappa [Kw] ⫽ 1.00 for both). The
reliability of scoring specific SI joint lesions and agreement
between the grade of the SI joint lesion was almost perfect
(36), with an overall exact agreement of 93% (Kw ⫽ 0.95).
Identification of chronic SI joint lesions on T1 scans had
excellent agreement, with 100% agreement (K ⫽ 1.00) for both
erosion and ankylosis. In the spine, agreement between different grades of different types of lesions (36) was also substantial
(prevalence-adjusted bias-adjusted kappa, 0.68 for RLs, 0.90
for end-plate lesions, 0.89 for posterior element lesions, 0.83
for spinous process lesions, and 0.90 for diffuse edema).
Statistical analysis. Cross-table analysis was used to
calculate the sensitivity, specificity, and likelihood ratios (LRs)
of specific lesions in relation to specific diagnoses, using the
treating physician’s diagnosis as the gold standard. LRs are
regarded as one of the best ways to measure and express
diagnostic accuracy (37). All statistical analyses were carried
out in SPSS, version 15.0.1.1 (SPSS, Chicago, IL).
RESULTS
Characteristics of the subjects. One hundred
eighty-five participants were recruited. The goldstandard physician diagnosis was SpA in 64 patients
(35%), DA in 45 (24%), and metastases in 45 (24%),
while 11 patients (6%) had normal clinical findings and
20 (11%) were diagnosed as having other conditions,
including tuberculosis abscess, multiple hemangiomas,
osteoporotic fractures, and radiation osteonecrosis. The
mean ⫾ SD age of the whole cohort was 52.5 ⫾ 17.3
years, ranging from 19 years to 91 years, of whom 89
(48%) were under age 50 years (47 patients with SpA, 18
with DA, 9 with malignancy, 4 with other diagnoses, and
11 with normal findings) and 98 (53%) were in the age
range of 30–60 years. As expected, SpA patients were
significantly younger (mean ⫾ SD 41.3 ⫾ 12.7 years)
than those with a gold-standard diagnosis of DA (61.9 ⫾
14.3 years; P ⬍ 0.05) or spinal malignancy (64.4 ⫾ 13.1
years; P ⬍ 0.05).
Imaging diagnosis versus gold-standard diagnosis. When all lesions were collated and a global
impression assigned, there was 72% agreement between
1334
Figure 1. Characteristic spondylarthritis (SpA) Romanus lesions
(RLs) in different diagnoses. A, Grade 3 RLs in the inferior-anterior
corner of T9 and grade 2 RLs in the anterior-superior of T10 (arrow
for both) in a patient with SpA. B, Grade 2 RLs in the inferior-anterior
of L3 (arrow) and a small annular tear in the L3–L4 disc in a patient
with degenerative spinal disease. C, Grade 2 RLs in the posteriorinferior corner of L3 (arrow) in a patient with metastatic breast cancer.
the imaging diagnosis and gold-standard diagnosis. The
agreement for a diagnosis of SpA alone was 75%, while
for DA it was 62%. The agreement for malignancy was
highest, at 91%. The most common discrepancy in
diagnosis was between SpA and DA, since both conditions commonly had RLs, end-plate lesions, and degenerative discs present. Degenerative discs were a common
finding across the whole cohort (38%), and therefore
this feature was of little diagnostic utility.
Incorrect diagnosis based on MRI findings
alone. In the SpA group, 16 subjects (25%) with a
gold-standard physician diagnosis of SpA in the study
(all fulfilling the ESSG criteria) were given another
diagnosis on the basis of the MRI scan alone. Twelve of
the 16 subjects were misdiagnosed as having DA (mean
age 41 years), based on MRI evidence of RLs (6 of 12),
end-plate lesions (9 of 12), and degenerative discs (9 of
12). These lesions were mostly mild and frequently
occurred in isolation, often adjacent to degenerative
discs in the lower cervical and lumbar spine. Three
subjects had normal findings on the MRI scans despite
active disease, which is consistent with cases described in
previous reports (10,13), and 1 patient was misdiagnosed
as having malignancy based on the MRI scans.
In the DA group, 17 patients were given a
diagnosis of SpA based on MRI alone. The median age
BENNETT ET AL
of this group was 59 years (range 26–79 years), and 88%,
53%, and 29% of the patients had RLs, end-plate lesions,
and posterior element lesions present, respectively.
The trend, both in SpA patients and in DA
patients, was for an increase in the number of BME
lesions with increasing age. The median lesion count per
SpA patient was 4.86 in those under age 50 years (n ⫽
44) and 5.5 in those over age 50 years (n ⫽ 19). In the
DA patients, the median lesion count was 3.3 in patients
under age 50 years (n ⫽ 12) and 5.2 in those over age 50
years (n ⫽ 33).
In the spinal malignancy group, RLs were frequently present on MRI; however, these were rarely
found in isolation and almost all cases also had other
abnormalities, such as expansile, lytic, or sclerotic lesions
in a single or multiple vertebrae or adjacent soft tissues,
clearly suggesting a diagnosis of malignancy. Nevertheless, 3 patients with confirmed metastatic disease (all
secondary to breast carcinoma) were given an imaging
diagnosis of SpA because of the limited distribution of
early metastatic disease at a stage when it was still
nondestructive (Figures 1C and 2A).
Figure 2. Incorrect magnetic resonance imaging (MRI) diagnoses
made on the basis of characteristic spondylarthritis (SpA) lesions
evident on MRI. A, Imaging diagnosis of SpA on the basis of Romanus
lesions (RLs) (grade 2) in the anterior-inferior border of T4 and
anterior-superior border of T5 (arrow for both), spinous process bone
marrow edema lesions at T3–T5 (grades 3, 2, and 2, respectively)
(triple arrow), and diffuse vertebral body edema at T9 (grade 2)
(arrowhead). The actual diagnosis is spinal bone metastases. B,
Imaging diagnosis of SpA on the basis of classic MRI evidence of
coalescent RLs anterior-superior and anterior-inferior to L4 (grade 1)
(arrow) on 3 consecutive slices, adjacent to well-hydrated healthy
intervertebral discs. The actual diagnosis is degenerative arthritis
(DA). C, Imaging diagnosis of DA on the basis of a large RL (grade 3)
anterior-inferior to L5 (arrow) in a classic “degenerative” location at
L5–S1, adjacent to a degenerative disc with a small anterior annular
tear. The actual diagnosis is psoriatic arthritis/SpA.
SPINAL MRI DIAGNOSIS OF AXIAL SpA
1335
Table 1. Sensitivity, specificity, and positive likelihood ratio (LR) for different grades and amounts of Romanus lesions (RLs) and different grades
of posterior element lesions in the diagnosis of spondylarthritis (SpA), degenerative spinal arthritis (DA), and spinal malignancy*
MRI characteristic, diagnosis
Sensitivity (95% CI)
Specificity (95% CI)
0.67 (0.53–0.79)
0.63 (0.47–0.77)
0.39 (0.25–0.54)
0.56 (0.47–0.65)
0.52 (0.43–0.61)
0.44 (0.35–0.53)
1.5 (1.2–2.0)
1.3 (1.0–1.8)
0.7 (0.5–1.0)
0.45 (0.32–0.59)
0.37 (0.23–0.53)
0.11 (0.04–0.25)
0.81 (0.73–0.88)
0.76 (0.67–0.83)
0.67 (0.58–0.75)
2.5 (1.5–4.0)†
1.5 (0.9–2.5)
0.4 (0.2–0.8)
0.33 (0.21–0.47)†
0.05 (0.01–0.18)
0.00 (0.00–0.10)
0.97 (0.92–0.99)†
0.85 (0.78–0.91)
0.83 (0.75–0.89)
12.4 (3.8–40.4)†
0.3 (0.1–1.4)
0.00
0.16 (0.08–0.29)
0.12 (0.05–0.27)
0.18 (0.09–0.33)
0.89 (0.81–0.94)
0.87 (0.79–0.92)
0.89 (0.82–0.94)
1.4 (0.7–3.2)
0.9 (0.4–2.3)
1.6 (0.7–3.6)
0.11 (0.05–0.23)
0.00 (0.00–0.11)
0.00 (0.00–0.10)
1.00 (0.96–1.00)
0.95 (0.90–0.98)
0.95 (0.89–0.98)
0.29 (0.18–0.43)
0.12 (0.05–0.27)
0.20 (0.10–0.36)
0.87 (0.79–0.92)
0.80 (0.71–0.86)
0.82 (0.74–0.88)
2.2 (1.2–4.1)
0.6 (0.3–1.5)
1.2 (0.6–2.3)
0.16 (0.08–0.29)
0.10 (0.03–0.24)
0.20 (0.10–0.36)
0.88 (0.80–0.93)
0.85 (0.78–0.91)
0.89 (0.82–0.94)
1.3 (0.6–2.9)
0.7 (0.2–1.8)
1.8 (0.9–3.9)
0.13 (0.06–0.25)
0.02 (0.00–0.14)
0.00 (0.00–0.10)
0.99 (0.94–1.00)
0.95 (0.89–0.98)
0.94 (0.87–0.97)
14.5 (1.8–115.0)†
0.5 (0.1–3.5)
0.0
0.33 (0.21–0.47)
0.66 (0.49–0.79)
0.36 (0.23–0.52)
0.59 (0.49–0.68)
0.70 (0.61–0.78)
0.61 (0.52–0.69)
0.8 (0.5–1.2)
2.2 (1.6–3.1)†
0.9 (0.6–1.5)
0.51 (0.37–0.64)
0.17 (0.08–0.33)
0.14 (0.06–0.28)
0.86 (0.78–0.92)
0.71 (0.62–0.79)
0.70 (0.61–0.77)
3.6 (2.2–6.1)†
0.6 (0.3–1.2)
0.4 (0.2–1.0)
Any RL (n ⫽ 87)
SpA
DA
Malignancy
ⱖ3 RLs (n ⫽ 46)
SpA
DA
Malignancy
ⱖ3 RLs and age ⱕ50 years (n ⫽ 21)
SpA
DA
Malignancy
Grade 3/severe RLs (n ⫽ 22)
SpA
DA
Malignancy
Grade 3/severe RLs and age ⱕ50 years (n ⫽ 6)
SpA
DA
Malignancy
Any posterior element lesion (n ⫽ 31)
SpA
DA
Malignancy
Grade 3 posterior element lesion (n ⫽ 23)
SpA
DA
Malignancy
Grade 1 or 2 posterior element lesions (n ⫽ 8)
SpA
DA
Malignancy
Any end-plate lesion (n ⫽ 65)
SpA
DA
Malignancy
Age ⱕ50 years and any RL or any posterior element lesion
(n ⫽ 44)
SpA
DA
Malignancy
Positive LR (range)
Infinite†
0.0
0.0
* MRI ⫽ magnetic resonance imaging; 95% CI ⫽ 95% confidence interval.
† Significant difference (P ⬍ 0.05) versus the other diagnoses.
Sacroiliac joint imaging. SI joint scans were
available for 89 subjects (48% of the total cohort).
Evaluation of the SI joints was included to determine
whether its inclusion in a spinal protocol aided in the
diagnosis. Fifty-six (63%) of the subjects with SI joint
scans were given a diagnosis of SpA and 11 (12%) had
normal findings, while the remaining SI joint scans
resulted in a diagnosis of degenerative disease in 9
subjects (10%), metastatic disease in 5 (5.6%), and other
diagnoses, including abscess and fracture, in 8 (8.9%). In
the SpA cohort, all of whom had clinically active disease,
24% of patients had normal findings on the SI joint
scans.
In 9 (10%) of the patients who had both spinal
and SI joint scans, the diagnosis of SpA was correctly
established on the basis of the SI joint scan when
findings on the spinal scans alone were equivocal. In a
further 9 patients (10%), the SI joint scan improved the
certainty of a diagnosis of SpA. High-signal STIR lesions
in the SI joints of patients with malignancy were often
different from those in the SI joints of patients with SpA,
with the SI joint scans in those with malignancy having a
more heterogeneous, “peppered” appearance. Patients
with DA had less frequent and less severe BME lesions
on STIR-sequence images of the SI joints than did SpA
patients, which supports the findings from previous
1336
Figure 3. Classic appearances of the spine on magnetic resonance
imaging (MRI) STIR sequences. A, A degenerative thoracolumbar
spine, showing widespread dehydrated extruding intervertebral discs
with marked end-plate edema at L2–L3 (short arrow) and a Romanus
lesion (RL) anterior-superior to T12 (long arrow). B, A spine with
spondylarthritis, showing multiple RLs at T3–T5 (arrows) with wellpreserved intervertebral discs. C, Spinal metastases, showing marked
bone destruction and expansile mass at T12 (short arrow), diffuse
vertebral body edema at T9 (arrowhead), an RL anterior-superior to
T8 (long arrow), and widespread patchy edema at L1–L5.
reports (38,39). Two control subjects with no back pain
each had a mild lesion evident in the SI joints.
Analysis of individual lesions. To explore the
global diagnosis findings in more detail, we undertook
analyses of different lesions.
Romanus lesions. RLs were the most frequently
observed lesion (n ⫽ 297 RLs) and were most commonly
found in the SpA group (67%, compared with 63% in
the DA group and 39% in the malignancy group).
However, the diagnostic utility of RLs (all grades of
severity included) for SpA was low and not significantly
better than for DA or spinal malignancy (Table 1).
Other features of malignancy, such as diffuse patchy
uptake with low signal on T1 and medium/high signals
on STIR, bone destruction, expansile/lytic/sclerotic lesions, or soft-tissue masses, generally allowed a correct
diagnosis of spinal malignancy to be made. However,
correctly diagnosing SpA while differentiating it from
DA was more difficult, because RLs also occurred frequently in DA (Figure 3A) and were not necessarily related
to obvious degenerative discs (Figure 2B).
BENNETT ET AL
The diagnostic utility of RLs for SpA, however,
improved with increasing severity of the RLs, with
increasing total number of RLs, and with younger age
(Table 1). When ⱖ3 RLs (any grade) was considered the
minimum cutoff value, the specificity and LR for SpA
increased to 0.81 and 2.5, respectively (Table 1). For
patients ⱕ50 years old with ⱖ3 (any grade) RLs, the
specificity and LR for SpA further increased significantly (P ⬍ 0.05 versus the other diagnoses), to 0.97 and
12.4, respectively. When only severe (grade 3) RLs in
patients ⱕ50 years were included, the specificity for SpA
increased further still, to 1.00, with an infinite positive
LR (P ⬍ 0.05 versus the other diagnoses). Both presence
of ⱖ3 (any grade) RLs and severe RLs in those ages ⱕ50
years had a large LR for SpA (LR ⬎10) (Table 1), which
represents substantial, and often conclusive, evidence of
this disease (40) and indicates that the presence of these
lesions in younger patients gives great accuracy to the
diagnosis of SpA.
Interestingly, the specificity of RLs for SpA was
not significantly improved in the absence of degenerative discs (in those with any RLs, sensitivity for SpA 0.68
[range 0.49–0.83], specificity for SpA 0.62 [range 0.50–
0.73]). Moreover, the presence of degenerative discs did
not improve the specificity of RLs for degenerative
spinal disease (in those with any RLs, sensitivity for DA
0.67 [range 0.45–0.84], specificity for DA 0.45 [range
0.30–0.61]), suggesting that severe RLs are a specific
feature of SpA regardless of the presence of degenerative disc disease. Nevertheless, RLs in DA were frequently, but not always, seen adjacent to classic areas of
degenerative disc disease in the lower cervical and
lumbar spine.
Posterior element and spinous process lesions. Posterior element BME lesions were present in 29% of SpA
patients, 20% of those with spinal malignancy, and 12%
of DA patients. Interestingly, these posterior element
lesions were almost exclusively severe (grade 3) in spinal
malignancy, but a more even distribution, between grade
1 and grade 3, was observed in SpA and DA. Any grade
of posterior element lesion had a high specificity for
SpA (0.87), although this was not significantly different
from the specificity for DA (0.80) or spinal malignancy
(0.82). However, if only mild (grade 1) or moderate
(grade 2) lesions were present, the specificity of posterior element lesions for SpA increased (specificity 0.99,
LR 14.5), and mild/moderate lesions were significantly
more specific for SpA than for DA or spinal malignancy
(P ⬍ 0.05). Severe posterior element lesions in patients
with malignancy were often associated with other char-
SPINAL MRI DIAGNOSIS OF AXIAL SpA
Figure 4. Posterior element lesions on magnetic resonance imaging
(MRI) STIR sequences from patients with spondylarthritis (SpA) and
patients with malignancy. A, Spine of a patient with ankylosing
spondylitis (AS), showing a posterior element/costovertebral grade 2
lesion at T12 (arrow). B, Spine of a patient with AS, showing a grade
1 lesion in the superior facet of T6 (short arrow) and grade 1 end-plate
lesion (long arrow) in the inferior end plate at T3. C, Spine of a patient
with metastatic breast cancer, showing posterior element grade 3
lesions at T9 and T10 (long arrow) and associated grade 3 diffuse
vertebral body edema at T9–T10 and bone destruction at T12 (short
arrow).
acteristic lesions of malignancy, such as bone destruction, as seen in Figure 4C.
Of course, the age of the patient is known during
routine interpretation of MRI, unlike the blinded procedures for this study. Thus, inclusion of age ⱕ50 years
in combination with any RL or any posterior element
lesion yielded a significantly greater LR for SpA (LR
3.6) than for DA (LR 0.58; P ⬍ 0.05) or spinal malignancy (LR 0.40; P ⬍ 0.05) (Table 1). Spinous process
involvement was seen more frequently in malignancy
(27%) and SpA (22%) as compared with DA (10%), but
only small, nonsignificant differences in sensitivity and
specificity were seen for spinous process BME lesions.
Diffuse vertebral body edema. Diffuse vertebral
BME was seen in patients with SpA (11%) and in those
with degenerative discs (22%) and malignant spines
(55%). Diffuse vertebral body edema (any grade/any
severity) had a significantly (P ⬍ 0.05) greater specificity
for spinal malignancy (specificity 0.84 [range 0.76–0.90],
LR 3.44) than for SpA (specificity 0.67 [range 0.57–
0.75], LR 0.33). The specificity and LR of diffuse body
edema for spinal malignancy increased further with
severe (grade 3) edema only (specificity 0.91 [range
0.84–0.95], LR 4.5; P ⬍ 0.05 versus the other diagnoses).
1337
End-plate edema. End-plate edema was common
in the 3 major diagnoses, but more so in DA (66%) than
in SpA (33%) or in spinal malignancy (36%). The LR of
end-plate edema for DA (LR 2.2) was significantly
greater than that for SpA (LR 0.8; P ⬍ 0.05) or that for
malignancy (LR 0.9; P ⬍ 0.05) (Table 1). End-plate
edema in DA was most frequently seen in the lumbar
spine, followed by the cervical spine and thoracic spine
(62%, 24%, and 14%, respectively). The lesions were
often (63% of end-plate lesions), but not always, seen at
classic degenerative sites (L3–L4, L4–L5, L5–S1, and
C4–C5, C5–C6, and C6–C7), with L5–S1 accounting for
20% of all end-plate lesions.
Location of lesions in SpA. Typical high-signal,
STIR-sequence BME (inflammatory) lesions in either
the spine or SI joints were seen in 61 (95%) of the 64
SpA patients. The remaining 3 SpA patients with active
disease (5%) had normal MRI findings on whole-spine
and SI joint scans. Acute SI joint lesions were present in
62% of SpA patients (a further 14% had chronic erosive
or ankylosed SI joints, without acute lesions), and 82%
of SpA patients had acute spinal lesions. The spinal
lesions were located in the T-spine in 62%, L-spine in
59%, and C-spine in 21% of SpA patients.
Despite having clinically active disease, 24% of
SpA patients had normal findings on the SI joint scans
and 18% had no acute lesions on the spinal scans.
Conversely, 75% of DA patients (6 of 8) had normal SI
joints, but only 9% (4 of 45) had no BME spinal lesions.
In patients with normal SI joints but abnormal findings
on the spine scans, acute spinal lesions were seen more
commonly in those with DA (mean 5.3 lesions per
patient) compared with those with SpA (mean 3.3
lesions per patient). In contrast, RLs were more frequent in SpA (61%) than in DA (47%), and severe
(grade 3) lesions were more common in SpA (16%)
compared with DA (0%). Importantly, in the SpA
cohort, 10% of patients had lesions in the cervical or
thoracic spine without associated lesions in the lumbar
spine of the SI joints.
DISCUSSION
This is the first study to systematically investigate
the diagnostic utility of characteristic SpA lesions on
MRI (5,41) in a cohort of patients with back pain with
mixed pathologic features. We showed that MRI assessment alone, without any clinical details, had good diagnostic utility in 3 causes of back pain, SpA, DA, and
malignancy, with 72% accuracy for all 3 diagnoses and
75% accuracy for SpA. When the MRI lesions were
1338
BENNETT ET AL
Figure 5. Characteristic spondylarthritis (SpA) lesions in patients with SpA and patients with malignancy. A,
Presence of diffuse vertebral body edema (arrow) and a Romanus lesion (RL) (arrowhead) in selected
thoracic vertebrae, both consistent with a diagnosis of SpA. B, Extended spinal views of the same patient as
in A, showing further features of SpA (spinous process edema [arrow]). C, Presence of an RL (arrow) and
diffuse vertebral body edema (arrowhead), consistent with a diagnosis of SpA, in selected thoracic vertebrae
of a patient. D, Extended spinal views of the same patient as in C, showing obvious features of malignant
disease with bone destruction and a soft-tissue mass at T12 (arrow).
interpreted in clinical context, especially according to
the patient’s age, and in detail, they were found to be
highly diagnostic of SpA. These findings could help in
establishing a definite diagnosis of axial SpA in the 23%
of cases in which the SI joints are spared (13). However,
the study also shows that typical SpA lesions are in fact
also common in other causes of back pain. In older
patients with later disease presentations, particularly in
those in whom only a few MRI lesions are present, the
findings indicate that considerable caution is needed
when ascribing a diagnosis.
We found that RLs are common in SpA, but
also are common in degenerative and malignant disease.
However, our detailed analysis of RLs indicated that the
presence of numerous or severe RLs in subjects ages
ⱕ50 years has high diagnostic accuracy for SpA. Although patients with malignant spinal disease also often
had RLs present and could be misdiagnosed as having
SpA at a vertebral level, this was in fact a very uncommon misdiagnosis at the patient level, since the overall
MRI picture was clearly indicative of malignant disease,
due to spinal imaging features elsewhere (Figure 5). We
have previously noted that the MRI pattern of disease in
the DIP joints in patients with PsA and those with OA is
often similar (14). Recent findings from a non–fat-
suppressed MRI study also suggested some potential
overlap in chronic changes between SpA and spinal OA
(42). Therefore, the presence of RLs on MRI in patients
with DA indicates that MRI findings of such lesions in
SpA must be carefully interpreted.
This study shows that although posterior element lesions on high-signal STIR images were less
frequent than RLs in SpA, mild and moderate posterior
element lesions were highly specific, having an extremely
high LR for a diagnosis of SpA. In contrast, severe posterior element lesions were suggestive of malignancy,
especially in older patients, and other MRI and clinical
signs of malignancy should be sought and excluded
before a diagnosis of SpA is entertained. This study was
not designed to develop a new spinal scoring system but
to describe the frequency, location, and morphology of
spinal lesions seen in SpA and other diseases, as occurs
in clinical practice, and to investigate the diagnostic
utility of these lesions. We believe that the high specificity of posterior element lesions for SpA implies that
the addition of posterior element assessment in existing
scoring systems might further improve their utility.
We found that degenerative discs on MRI were
of low diagnostic utility when trying to distinguish
between back pain in SpA and that in DA. It has
SPINAL MRI DIAGNOSIS OF AXIAL SpA
previously been accepted that RLs or end-plate lesions
occurring on either side of degenerative discs are likely
to be degenerative in nature (31,35). However, in this
study, degenerative discs were common in normal subjects, SpA patients, and patients with spinal malignancy,
as well as in patients with DA. We recognize that the
present findings come from a cohort of SpA patients
whose mean age was 41 years and whose mean disease
duration was 8.5 years, and that the specificity of degenerative discs for DA would probably have been higher
if the studied subjects were under age 30 years. However, we specifically aimed to study a wide age range to
capture all the possible ages at onset of SpA (16–18).
In particular, we were interested in studying the 30–
50-year-old age group, since we believe that these subjects may provide the most difficult diagnostic imaging
dilemmas because of the common acute inflammatorylike MRI lesions that occur in degenerative spines in this
age group (19) and which may be mistaken for SpA. Our
findings confirmed that end-plate lesions had a significantly higher LR for a diagnosis of DA. However, this
finding should be interpreted with caution, since endplate lesions were also common in SpA and malignancy.
The high diagnostic utility of the above-described
RLs and posterior element lesions for SpA could be of
considerable diagnostic importance, since chronic back
pain is often investigated by nonrheumatologists and SI
joint scanning may be excluded. Another important
consideration is which spinal regions to scan. In most
cases of suspected SpA, lumbar spine and SI joint MRI
sequences are clinically investigated. However, acute
BME lesions in SpA were most frequently seen in the
thoracic spine (62%), which is consistent with previously
reported data in patients with AS (43). In addition, 10%
of SpA patients in the present cohort had only thoracic
or cervical spine lesions, and 24% had normal findings
on their SI joint scans. This indicates that scanning the
whole spine as well as the SI joints may be a more
diagnostically useful tool when investigating potential
SpA. Also of interest is that 5% of SpA patients had no
acute inflammatory lesions on MRI STIR sequences or
on scans of the whole spine and SI joints, despite having
clinically active disease.
This study has some limitations. Only sagittal
images of the spine were scored; therefore, some lesions
that may be evident only on coronal planes may have
been missed. Also, MRI findings were evaluated and the
most probable single diagnosis was given based on these
findings; therefore, the possibility of coexisting conditions was not taken into account. Finally, the present
SpA cohort generally had established disease, with only
1339
a minority having early disease. Nevertheless, the evidence suggests that the MRI appearances of early SpA
are similar to those of late disease (3,38), and studying a
cohort with a mean disease duration of 8.5 years may
have captured a time point early enough for up to 40%
of subjects to not have developed diagnostic radiographic features (44). In this scenario, specific MRI
lesions would be of great assistance in making diagnoses.
Our study has thus shown that MRI has good
diagnostic utility in SpA, with a 75% rate of correct
diagnosis made by scan alone. However, characteristic
SpA lesions on MRI are also frequently present in those
with malignant, and particularly degenerative, spinal
disease. This may lead to misdiagnosis, especially in
older people, and therefore the findings need to be
examined in closer detail. Severe or multiple RLs on
MRI in younger patients and mild/moderate posterior
element lesions are highly diagnostic of SpA, irrespective of any other clinical or laboratory parameter.
AUTHOR CONTRIBUTIONS
All authors were involved in drafting the article or revising it
critically for important intellectual content, and all authors approved
the final version to be published. Dr. Bennett and Prof. McGonagle
had full access to all of the data in the study and take responsibility for
the integrity of the data and the accuracy of the data analysis.
Study conception and design. Bennett, Rehman, Hensor, MarzoOrtega, McGonagle.
Acquisition of data. Bennett, Rehman, Hensor, Marzo-Ortega,
McGonagle.
Analysis and interpretation of data. Bennett, Rehman, Hensor,
Marzo-Ortega, McGonagle.
REFERENCES
1. Mau W, Zeidler H, Mau R, Majewski A, Freyschmidt J, Stangel
W, et al. Clinical features and prognosis of patients with possible
ankylosing spondylitis: results of a 10-year follow-up. J Rheumatol
1988;15:1109–14.
2. Puhakka KB, Jurik AG, Schiottz-Christensen B, Hansen GV,
Egund N, Christiansen JV, et al. MRI abnormalities of sacroiliac
joints in early spondylarthropathy: a 1-year follow-up study. Scand
J Rheumatol 2004;33:332–8.
3. Bennett AN, McGonagle D, O’Connor P, Hensor EM, Sivera F,
Coates LC, et al. Severity of baseline magnetic resonance
imaging–evident sacroiliitis and HLA–B27 status in early inflammatory back pain predict radiographically evident ankylosing
spondylitis at eight years. Arthritis Rheum 2008;58:3413–8.
4. Benjamin M, McGonagle D. The anatomical basis for disease
localisation in seronegative spondyloarthropathy at entheses and
related sites. J Anat 2001;199(Pt 5):503–26.
5. Maksymowych WP, Lambert RG. Magnetic resonance imaging for
spondyloarthritis: avoiding the minefield. J Rheumatol 2007;34:
259–65.
6. Marzo-Ortega H, McGonagle D, O’Connor P, Emery P. Efficacy
of etanercept in the treatment of the entheseal pathology in
resistant spondylarthropathy: a clinical and magnetic resonance
imaging study. Arthritis Rheum 2001;44:2112–7.
7. Marzo H, McGonagle D, O’Connor P, Green MJ, Pease C,
1340
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
Wakefield R, et al. Fat suppressed MRI in spinal disease in early
spondyloarthropathy [abstract]. Arthritis Rheum 1998;41 Suppl
9:S355.
Rudwaleit M, van der Heijde D, Khan MA, Braun J, Sieper J. How
to diagnose axial spondyloarthritis early. Ann Rheum Dis 2004;63:
535–43.
Maksymowych WP, Inman RD, Salonen D, Dhillon SS, Krishnananthan R, Stone M, et al. Spondyloarthritis Research Consortium
of Canada magnetic resonance imaging index for assessment of
spinal inflammation in ankylosing spondylitis. Arthritis Rheum
2005;53:502–9.
Braun J, Baraliakos X, Golder W, Brandt J, Rudwaleit M, Listing
J, et al. Magnetic resonance imaging examinations of the spine in
patients with ankylosing spondylitis, before and after successful
therapy with infliximab: evaluation of a new scoring system.
Arthritis Rheum 2003;48:1126–36.
Rudwaleit M, Schwarzlose S, Listing J, Brandt J, Braun J, Sieper
J. Is there a place for magnetic resonance imaging (MRI) in
predicting a major clinical response (BASDAI 50) to TNF alpha
blockers in ankylosing spondylitis [abstract]. Arthritis Rheum
2004;50 Suppl 9:S211.
Marzo-Ortega H, McGonagle D, Jarrett S, Haugeberg G, Hensor
E, O’Connor P, et al. Infliximab in combination with methotrexate
in active ankylosing spondylitis: a clinical and imaging study. Ann
Rheum Dis 2005;64:1568–75.
Rudwaleit M, Schwarzlose S, Hilgert ES, Listing J, Braun J, Sieper
J. MRI in predicting a major clinical response to anti-tumour
necrosis factor treatment in ankylosing spondylitis. Ann Rheum
Dis 2008;67:1276–81.
Tan AL, Grainger AJ, Tanner SF, Emery P, McGonagle D. A
high-resolution magnetic resonance imaging study of distal interphalangeal joint arthropathy in psoriatic arthritis and osteoarthritis: are they the same? Arthritis Rheum 2006;54:1328–33.
Dougados M, van der Linden S, Juhlin R, Huitfeldt B, Amor B,
Calin A, et al, and the European Spondylarthropathy Study
Group. The European Spondylarthropathy Study Group preliminary criteria for the classification of spondylarthropathy. Arthritis
Rheum 1991;34:1218–27.
Olivieri I, Salvarani C, Cantini F, Ciancio G, Padula A. Ankylosing
spondylitis and undifferentiated spondyloarthropathies: a clinical
review and description of a disease subset with older age at onset.
Curr Opin Rheumatol 2001;13:280–4.
Shankaranarayana S, Tan AL, Madden J, McGonagle D.
Fat-suppression magnetic resonance imaging in the diagnosis of
late-onset axial spondyloarthropathy. Ann Rheum Dis 2007;66:
1405–6.
Toussirot E, Wendling D. Late-onset ankylosing spondylitis and
related spondylarthropathies: clinical and radiological characteristics and pharmacological treatment options. Drugs Aging 2005;
22:451–69.
Kuisma M, Karppinen J, Niinimaki J, Ojala R, Haapea M,
Heliovaara M, et al. Modic changes in endplates of lumbar
vertebral bodies: prevalence and association with low back and
sciatic pain among middle-aged male workers. Spine 2007;32:
1116–22.
Van der Heijde DM, Landewe RB, Hermann KG, Jurik AG,
Maksymowych WP, Rudwaleit M, et al. Application of the
OMERACT filter to scoring methods for magnetic resonance
imaging of the sacroiliac joints and the spine: recommendations
for a research agenda at OMERACT 7. J Rheumatol 2005;32:
2042–7.
Carley S, Dosman S, Jones SR, Harrison M. Simple nomograms
to calculate sample size in diagnostic studies. Emerg Med J 2005;
22:180–1.
Hermann KG, Landewe RB, Braun J, van der Heijde DM.
Magnetic resonance imaging of inflammatory lesions in the spine
BENNETT ET AL
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
in ankylosing spondylitis clinical trials: is paramagnetic contrast
medium necessary? J Rheumatol 2005;32:2056–60.
Ostergaard M, Edmonds J, McQueen F, Peterfy C, Lassere M,
Ejbjerg B, et al. An introduction to the EULAR–OMERACT
rheumatoid arthritis MRI reference image atlas. Ann Rheum Dis
2005;64 Suppl I:i3–7.
Landewe RB, Hermann KG, van der Heijde DM, Baraliakos X,
Jurik AG, Lambert RG, et al. Scoring sacroiliac joints by magnetic
resonance imaging: a multiple-reader reliability experiment.
J Rheumatol 2005;32:2050–5.
Maksymowych WP, Dhillon SS, Park R, Salonen D, Inman RD,
Lambert RG. Validation of the Spondyloarthritis Research Consortium of Canada magnetic resonance imaging spinal inflammation index: is it necessary to score the entire spine? Arthritis
Rheum 2007;57:501–7.
Crowther SM, Lambert RG, Dhillon SS, Maksymowych WP. High
frequency of inflammatory lesions in the posterior structures of
the spine in patients with ankylosing spondylitis (AS): a systematic
evaluation by MRI [abstract]. Arthritis Rheum 2006;54 Suppl
9:S793.
Bochkova A, Lewschakova A, Bunchuk N. Inflammatory spinal
MRI lesions in patients with ankylosing spondylitis and short
history of axial pain prevail in the posterior structures. Ann
Rheum Dis 2008;67 Supp II:379.
Lukas C, Braun J, van der Heijde D, Hermann KG, Rudwaleit M,
Ostergaard M, et al. Scoring inflammatory activity of the spine by
magnetic resonance imaging in ankylosing spondylitis: a multireader experiment. J Rheumatol 2007;34:862–70.
Van der Heijde D, Landewe R, Hermann KG, Rudwaleit M,
Ostergaard M, Oostveen A, et al. Is there a preferred method for
scoring activity of the spine by magnetic resonance imaging in
ankylosing spondylitis? J Rheumatol 2007;34:871–3.
MacVicar D. Imaging of the spine in patients with malignancy.
Cancer Imaging 2006;6:S22–6.
Rankine JJ, Hutchinson CE, Hughes DG. MRI of lumbar spondylosis: a comparison of sagittal T2 weighted and three sequence
examinations. Br J Radiol 1997;70:1112–21.
McGonagle D, Gibbon W, O’Connor P, Green M, Pease C, Emery
P. Characteristic magnetic resonance imaging entheseal changes
of knee synovitis in spondylarthropathy. Arthritis Rheum 1998;41:
694–700.
McGonagle D, Marzo-Ortega H, O’Connor P, Gibbon W,
Pease C, Reece R, et al. The role of biomechanical factors and
HLA–B27 in magnetic resonance imaging–determined bone
changes in plantar fascia enthesopathy. Arthritis Rheum 2002;46:
489–93.
Haibel H, Rudwaleit M, Brandt HC, Grozdanovic Z, Listing J,
Kupper H, et al. Adalimumab reduces spinal symptoms in active
ankylosing spondylitis: clinical and magnetic resonance imaging
results of a fifty-two–week open-label trial. Arthritis Rheum
2006;54:678–81.
Karchevsky M, Schweitzer ME, Carrino JA, Zoga A, Montgomery
D, Parker L. Reactive endplate marrow changes: a systematic
morphologic and epidemiologic evaluation. Skeletal Radiol 2005;
34:125–9.
Landis JR, Koch GG. The measurement of observer agreement
for categorical data. Biometrics 1977;33:159–74.
McGee S. Simplifying likelihood ratios. J Gen Intern Med 2002;
17:646–9.
Marzo-Ortega H, McGonagle D, O’Connor P, Hensor EM, Bennett AN, Green M, et al. Baseline and one year magnetic
resonance imaging of the sacroiliac joints and lumbar spine in
early inflammatory back pain: the relationship between HLA-B27,
symptoms and disease extent and persistence. Ann Rheum Dis
2008;67 Sup II:ii377–8.
Rudwaleit M, Vahldiek J, Wenz J, Brandt H, Appel H, Haibel H,
et al. Sensitivity and specificity of magnetic resonance imaging
SPINAL MRI DIAGNOSIS OF AXIAL SpA
(MRI) of the sacroiliac joints in patients with suspected early
ankylosing spondylitis [abstract]. Arthritis Rheum 2006;54 Suppl
9:S793–4.
40. Jaeschke R, Guyatt GH, Sackett DL, the Evidence-Based Medicine Working Group. Users’ guides to the medical literature. III.
How to use an article about a diagnostic test. B. What are the
results and will they help me in caring for my patients? JAMA
1994;271:703–7.
41. Zochling J, Baraliakos X, Hermann KG, Braun J. Magnetic
resonance imaging in ankylosing spondylitis. Curr Opin Rheumatol 2007;19:346–52.
1341
42. Kim NR, Choi JY, Hong SH, Jun WS, Lee JW, Choi JA, et al.
“MR corner sign”: value for predicting presence of ankylosing
spondylitis. AJR Am J Roentgenol 2008;191:124–8.
43. Baraliakos X, Landewe R, Hermann KG, Listing J, Golder W,
Brandt J, et al. Inflammation in ankylosing spondylitis: a systematic description of the extent and frequency of acute spinal changes
using magnetic resonance imaging. Ann Rheum Dis 2005;64:
730–4.
44. Rudwaleit M, Khan MA, Sieper J. The challenge of diagnosis and
classification in early ankylosing spondylitis: do we need new
criteria? Arthritis Rheum 2005;52:1000–8.
DOI 10.1002/art.24455
Clinical Images: Cement embolism following percutaneous vertebroplasty
The patient, a 78-year-old woman, presented with back pain, which she had been experiencing for several months. Plain radiography
revealed a compression fracture of the first and second vertebral bodies and radiodense materials in the first vertebral body (left).
In addition, along the right border of the lumbar vertebrae was a radiopaque vascular structure, which seemed to be the ascending
lumbar vein (left). A radiograph of the chest showed branching, tubular, radiodense lines in the lungs (right). After further inquiry,
the patient revealed that she had undergone percutaneous vertebroplasty for an osteoporotic lumbar compression fracture 3 years
prior. Cement embolism due to the leakage of acrylic bone cement was diagnosed.
Hyeok-Jae Ko, MD
Kyung-Su Park, MD
Chul-Soo Cho, MD
Ho-Youn Kim, MD
The Catholic University of Korea College of Medicine
Seoul, Korea