Rheumatology 2015;54:1039–1049

RHEUMATOLOGY doi:10.1093/rheumatology/keu439
Advance Access publication 26 November 2014

Original article
Head-to-toe whole-body MRI in psoriatic
arthritis, axial spondyloarthritis and healthy
subjects: first steps towards global
inflammation and damage scores of

Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019

peripheral and axial joints
René Panduro Poggenborg1,2, Susanne Juhl Pedersen1,2,3, Iris Eshed4,
Inge Juul Sørensen1,2, Jakob M. Møller5, Ole Rintek Madsen3,
Henrik S. Thomsen5 and Mikkel Østergaard1,2

Abstract
Objectives. By whole-body MRI (WBMRI), we aimed to examine the frequency and distribution of inflam-
matory and structural lesions in PsA patients, SpA patients and healthy subjects (HSs), to introduce global
WBMRI inflammation/damage scores, and to assess WBMRI’s reproducibility and correlation with con-
ventional MRI (convMRI).
Methods. WBMRI (3.0-T) of patients with peripheral PsA (n = 18) or axial SpA (n = 18) and of HS (n = 12)
was examined for proportion of evaluable features (readability) and the presence and pattern of lesions in
axial and peripheral joints. Furthermore, global WBMRI scores of inflammation and structural damage
were constructed, and WBMRI findings were compared with clinical measures and convMRI (SpA/HS:
spine and SI joints; PsA/HS: hand).
Results. The readability (92–100%) and reproducibility (intrareader intraclass correlation coefficient:
0.62–1.0) were high in spine/SI joint, but lower in the distal peripheral joints. Wrists, shoulders, knees,

CLINICAL
SCIENCE
ankles and MTP joints were most commonly involved, with frequency of synovitis > bone marrow oedema
(BMO) > erosion. WBMRI global BMO scores of peripheral and axial joints were higher in PsA {median 7
[interquartile range (IQR) 3–15]} and SpA [8 (IQR 2–14)] than in HSs [2.5 (IQR 1–4.5)], both P < 0.05.
WBMRI global structural damage scores (erosion, fat infiltration and ankylosis) were higher in SpA
[7 (IQR 3–12)] than HSs [1.5 (IQR 0–4.5)], P = 0.012. Correlations between WBMRI and convMRI spine
and SI joint scores were r = 0.20–0.78.
Conclusion. WBMRI allows simultaneous assessment of peripheral and axial joints in PsA and SpA, and
the distribution of inflammatory and structural lesions and global scores can be determined. The study
strongly encourages further development and longitudinal testing of WBMRI techniques and assessment
methods in PsA and SpA.
Key words: psoriatic arthritis, spondyloarthritis, MRI, clinical examination, healthy subject.

1
Copenhagen Center for Arthritis Research, Copenhagen Center for
Rheumatology and Spinal Diseases, University Hospital Copenhagen
Glostrup, Copenhagen, 2Department of Clinical Medicine, Faculty of
Submitted 12 March 2014; revised version accepted
Health and Medical Sciences, University of Copenhagen, 3Department
17 September 2014
of Rheumatology and Medicine, University Hospital Copenhagen
Gentofte, Copenhagen, Denmark, 4Department of Diagnostic Imaging, Correspondence to: René Panduro Poggenborg, Copenhagen Center
Sheba Medical Center, Tel Giborim, Tel Aviv University Israel and for Rheumatology and Spinal Diseases (VRR), Glostrup Hospital,
5
Department of Radiology, University Hospital Copenhagen Herlev, Nordre Ringvej 57, DK-2600 Glostrup, Denmark.
Copenhagen, Denmark E-mail: poggenborg@dadlnet.dk

! The Author 2014. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oup.com
René Panduro Poggenborg et al.
Rheumatology key messages
. Whole-body MRI can visualize inflammatory and structural changes in PsA and SpA patients.
. Bone marrow oedema is more frequent in PsA and SpA patients than in healthy subjects.
. With further optimization, whole-body MRI could be a future tool for global disease assessment.
Introduction
In the management of PsA and axial SpA, it is important to
accurately monitor current disease activity and the
amount of structural damage [1]. MRI enables more
sensitive assessment of both inflammatory and structural
lesions than conventional clinical and radiographic
measures [2, 3]. However, both PsA and SpA are
characterized by diverse and anatomically widespread
disease manifestations, whereas conventional MRI
(convMRI) only allows visualization of a small anatomical
area at one examination, which limits its ability to
provide an overall assessment of disease status in these
diseases. Recently, head-to-toe whole-body MRI
(WBMRI) has been introduced. This method allows as-
sessment of all peripheral and axial joints, including
spine and SI joints, in only one examination [4, 5]. We
have previously reported findings of enthesitis using
WBMRI and clinical examination in PsA, SpA and healthy
subjects (HSs) [6]. Only a few studies have investigated
the ability of head-to-toe WBMRI to assess inflammation
and/or structural joint damage in patients with PsA [7] or
axial SpA [8–10]. None of these studies have compared
WBMRI findings of both peripheral and axial arthritis with
convMRI or have included healthy subjects (HSs).
Furthermore, none of the studies have assessed the pos-
sibility of developing WBMRI global inflammation
and structural damage scores including both peripheral
and axial joints. Thus, further development and validation,
including identification of findings in HSs is needed.
The perspective is the development of tools for assess-
ment of patient global inflammation and patient global
structural damage, capturing both peripheral and axial
features.
In this pilot study, we aimed to (i) examine the ability of
WBMRI to visualize inflammatory and structural lesions in
axial and peripheral joints, (ii) determine the distribution of
peripheral and axial involvement in PsA and SpA, (iii) val-
idate findings by comparison with clinical examination,
convMRI and by assessing the reproducibility, and (iv) de-
velop WBMRI global inflammation and global structural
damage scores.
Methods
Study participants
Patients were eligible for this prospective study if they
had PsA according to Moll and Wright’s criteria [11], or
SpA according to the European Spondylarthropathy
Study Group criteria [12]. PsA patients were included if
they had active disease, defined as a tender joint count
(TJC) or swollen joint count (SJC) of 53 and 51 swollen
finger joint or dactylitic finger. SpA patients were
included if they had a BASDAI [13, 14] score of 530 mm
1040
(0–100 mm scale) and active spinal disease according
to the treating rheumatologist. Treatment with gluco-
corticoids or initiation of TNFa inhibitor therapy was not
allowed within the 4 weeks before study investigations.
The study also included a group of HSs, who could
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
not be included if they had pain from peripheral
joints or spine, a family history of PsA, SpA or RA, a
medical history of psoriasis, anterior uveitis, IBD, or heel
pain.
The study was performed in accordance with the
Declaration of Helsinki and approved by the Ethical
Committee of The Capital Region of Denmark and the
Danish Data Protection Agency. Written informed consent
was obtained from all study participants before inclusion
into the study.
Clinical examinations
The clinical examination included the 76-SJC and 78-TJC
[15] and was performed by experienced rheumatologists.
BASDAI, BASFI and BASMI were acquired, as were serum
CRP, DAS28(CRP) (four variables).
MRI acquisition
All study participants had a head-to-toe WBMRI per-
formed. Patients with PsA also had convMRI of the
fingers, including second–fifth MCP, PIP and DIP joints.
Patients with axial SpA had convMRI of the SI joints
and total spine. The HSs had convMRI performed for
the fingers (n = 2) or spine (n = 8), or fingers and spine
(n = 2). WBMRI was performed on a 3T high-field MRI
unit (Philips Achieva, Best, The Netherlands) using
the integrated quadrature body coil. Pre-contrast short
tau inversion recovery and pre- and post-contrast
T1-weighted turbo spin echo sequences were performed
in six stations with coronal slice orientation (cervical,
thoracic, lumbar, hips/hands, knees and feet), one
station with sagittal (cervical) and one station with axial
orientation (feet). convMRI was performed on a 1.5T
high-field MRI unit (Philips Achieva, Best, The
Netherlands). Details on WBMRI and convMRI acquisition
can be found in Table 1. Scan time for WBMRI
was  60 min; scan time for convMRI 20 min for the
hands (PsA and HSs) and 40 min for the spine and SI
joints (SpA and HSs).
MRI evaluation
WBMRI and convMRI were evaluated separately by
an experienced musculoskeletal radiologist (I.E.), who
was blinded to all clinical, biochemical and other
imaging information and evaluated the images in random
order.
By WBMRI, the 76 joints (hereafter named the 76 per-
ipheral joints) included in the 78-TJC minus the TM joints,
www.rheumatology.oxfordjournals.org
www.rheumatology.oxfordjournals.org
TABLE 1 Whole-body MRI and conventional MRI imaging protocol for examination of patients with PsA and SpA and healthy subjects

Subjects Sequence Stations Orientation TR, ms TE, ms TI, ms FOV, mm Matrix Thk, mm Gap, mm Time, min:s

WBMRI (3T)
All T1 Cervical, thoracic, lumbar, knee, Coronal 733 7.6 — 470  253–259 312  171 5 0.5 2:15
Hips/hands, feet 1099 428  228 3 0.3 3:37
Feet Axial 2:15
Neck Sagittal 916 312  168 4 0.4 3:37
STIR Cervical, thoracic, lumbar, hips/ Coronal 6572 70 200 470  279–287 256  104 5 0.5 1:03
hands, knee, feet
Feet Axial
Cervical Sagittal 5258
Hips Coronal 13 905 83 380  213 3 0.3 4:38
Conventional
MRI (1.5T)
PsA + HS T1 Hand Coronal 40 12 — 120  120 150  150 0.8 0 5:01
STIR 4548 30 150 120  120 148  149 2 4:24
SpA + HS T1 Cervical, thoracic, lumbar Sagittal 400 7.4 — 270  270 300  216 4 0.4 2:59
SI joints Oblique coronal 550 14 300  270 320  171 2:55
STIR Cervical, thoracic, lumbar Sagittal 5000 80 120 270  270 272  216 0.8 4:10
SI joints Oblique coronal 2500 60 160 300  239 256  163 2:05

T1: T1-weighted sequences acquired before and after administration of gadoteric acid (Dotarem, Guerbet, 0.5 mmol/ml, 0.2 ml/kg). Conventional MRIs of the spine and SI joints are
without contrast. WBMRI: whole-body MRI; FOV: field of view; Gap: gap between slices; HS: healthy subjects; STIR: short tau inversion recovery; TE: echo time; TI: inversion time;
Thk: slice thickness; TR: repetition time.
Whole-body MRI in PsA and SpA

1041
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
René Panduro Poggenborg et al.
which were not visualized by WBMRI) were evaluated.
Furthermore, each discovertebral unit (DVU) in the spine
was evaluated separately in the ventral part (vertebral
body) and posterior part (including pedicles, facet
joints and spinous processes), whereas the SI joints
were evaluated on quadrant level, that is, by dividing
each SI joint into upper and lower iliac parts and
upper and lower sacral parts. The readability of WBMRI
was assessed for all inflammatory and structural lesions in
all scanned joint regions. Each joint region was classified
into one of three groups: (i) in the field of view (FOV) and
possible to evaluate (i.e. readable); (ii) in FOV, but not
possible to evaluate; and (iii) not in the FOV (i.e. not
scanned).
In readable areas, WBMRI was scored dichotomously
(presence/absence) for bone marrow oedema (BMO) and
bone erosions (in 76 peripheral joints, spine and SI joints),
synovitis (in 76 peripheral joints), fat infiltration (in spine
and SI joints) and ankylosis (in SI joints).
convMRI of the hands was scored for synovitis (0–3 per
joint), BMO (0–3 per bone) and bone erosion (0–10 per
bone), according to the Outcome Measure in the
Rheumatology PsA MRI scoring system PsAMRIS [16,
17]. convMRI of the spine was scored for activity (evalu-
ating BMO; 0–3 per DVU) using the Berlin modification of
the AS spine MRI activity score [18], and for structural
lesions (global score of sclerosis, squaring, erosion, syn-
desmophytes and bridging/fusion; 0–6 per DVU) using the
AS spine MRI-chronicity (ASspiMRI-c) score [19].
convMRI of the SI joint was scored for activity (evaluating
bone oedema; 0–4 per quadrant) and structural lesions
(global score of bone erosions, sclerosis, joint space
width, bone bridging/ankylosis; 0–1 per quadrant) using
the Berlin method [1, 20].
Construction of WBMRI scores including WBMRI
global scores
After assessment of all MRIs, different WBMRI peripheral
joint scores were calculated for synovitis, BMO and ero-
sion in the 76 peripheral joints. WBMRI spine scores were
calculated separately for BMO, fat infiltration and erosion
by adding scores for vertebral bodies and posterior
elements. For WBMRI, SI joint scores were calculated
separately for BMO, fat infiltration, erosion and ankylosis
by adding the scores from the quadrants. Furthermore,
different WBMRI global scores were calculated for BMO,
inflammation and structural damage, respectively. The
WBMRI global inflammation score was calculated as
the sum of synovitis and BMO in the 76 peripheral
joints, and BMO in the spine and SI joints. The WBMRI
global structural damage score was calculated as the sum
of erosions in 76 peripheral joints, spine and SI joints, fat
infiltrations in the spine and SI joints and ankylosis in the
SI joints.
Statistics
Results are given as median and interquartile range (IQR).
The Mann–Whitney test and Fisher’s exact test were used
to compare groups. Spearman’s rank correlation
1042
coefficient (rho) was used to assess correlations, and
values >0.20 but 40.50 and >0.50 but 40.80 and
>0.80 were considered to represent small, moderate
and high correlation, respectively. Reliability analyses
included intrareader intraclass correlation coefficients
(ICCs; two-way mixed model, absolute agreement defin-
ition). The statistical analyses were performed in SAS (ver-
sion 9. SAS Inc., Cary, NC, USA), except for ICCs, which
were performed in SPSS version 19 (SPSS Inc., Chicago,
IL, USA). P < 0.05 was considered statistically significant.
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
Results
Clinical findings
The study included 18 patients with PsA, 18 patients with
axial SpA, and 12 HSs. The characteristics of the study
participants are provided in Table 2. Compared with HSs,
patients with PsA and SpA had higher TJCs and SJCs,
DAS28(CRP), BASDAI and BASFI (0.0005 < P < 0.05).
Readability of WBMRI
Readability was examined in 76 peripheral joints, 23 spinal
DVUs and 8 SI joint quadrants (Table 3). Among the 76
peripheral joints evaluated, we found the highest readabil-
ity for synovitis, BMO and erosions in the knees
(94–100%) and hips (97–100%). Distal peripheral joints
generally showed lower readability than more proximal
joints; for example, BMO could only be evaluated in
10% of feet DIP joints, due to not being in the FOV
(37%), or due to insufficient image quality (53%). The
elbows could be evaluated in only 5–8%. In the spine
and SI joints, all areas showed high readability (592%).
WBMRI findings
Fig. 1 shows examples of inflammatory lesions detected
by WBMRI. The distribution of inflammatory and structural
changes in the 76 peripheral joints, spine and SI joints
examined are shown in Fig. 2.
In peripheral joints, synovitis was overall more frequent
than BMO and much more common than erosion. Hand
synovitis (particularly in PIP and DIP joints) was more fre-
quent in PsA than in SpA and HSs. In PsA, the frequency
of inflammatory involvement (synovitis and BMO) was
similar in hands and feet, whereas SpA patients had
more inflammatory lesions in feet than in hands. Large
joint involvement was frequent in both PsA and SpA, par-
ticularly in shoulders (synovitis in 48% and 42%, respect-
ively), wrists (48% and 50%), knees (42% and 17%) and
ankles (44% and 31%) (Table 3).
In the spine, BMO (SpA: 10% and PsA: 8% of vertebral
bodies) and fat infiltration (SpA: 12%; PsA: 7%) were the
most common findings, and both were more common in
SpA than in PsA, whereas erosions were rarely observed
(Table 3). BMO in the posterior parts of the spine was
observed occasionally in PsA (2% of sites) and SpA
(0.4%). In SI joints, BMO (SpA: 21% of quadrants; PsA:
4%) and fat infiltration (SpA: 31%; PsA 12%) were most
commonly observed, and BMO, fat infiltration, ankylosis
and erosions were more frequent in lower quadrants than
www.rheumatology.oxfordjournals.org
 Whole-body MRI in PsA and SpA

TABLE 2 Baseline characteristics of the study participants

PsA (n = 18) SpA (n = 18) Healthy subjects (n = 12)

Age, years 49 (37–58) 42 (32–51) 32 (27–47)

Sex, female, n (%) 11 (61) 8 (44) 8 (67)
Disease duration, years 3 (1–9) 4 (2–8) NA
SJC (0–76) 5 (3–11) 1 (0–2) 0 (0–0)
TJC (0–78) 13 (7–30) 4 (0–17) 0 (0–0)
DAS28 (CRP) 4.3 (3.4–4.6) 3.6 (1.9–3.9) 1.8 (1.5–1.9)
BASDAI (0–100 mm VAS) 44 (19–70) 56 (46–68) 2 (1–4)
BASFI (0–100 mm VAS) 19 (9–43) 46 (28–64) 0 (0–1)

Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019

BASMI (0–10) 1 (0–1) 3 (2–4) 0 (0–0)

Values are median (IQR) or frequency (%). Disease duration is years since diagnosis. NA: not applicable; SJC: swollen joint
count; TJC: tender joint count.

TABLE 3 Whole-body MRI readability and frequency of lesions in PsA and SpA

76 peripheral joints

Readable / in FOV, but not No. of lesions,

readable, %/%a percentage of readable areasb

Synovitis BMO Bone erosion

Synovitis BMO Bone erosion
All PsA SpA PsA SpA PsA SpA

Shoulders and AC joints (4) 93/0 90/0 92/0 32 (48) 27 (42) 19 (31) 15 (23) 0 (0) 5 (8)
Elbows (2) 8/0 6/0 5/0 3 (75) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
Wrists and first CMC joints (4) 84/0 83/1 86/0 33 (55) 29 (52) 14 (23) 10 (19) 6 (10) 5 (8)
MCP joints (10) 83/1 79/6 84/16 40 (28) 18 (13) 13 (10) 4 (3) 2 (1) 3 (2)
PIP joints (10) 65/9 59/14 69/8 20 (17) 8 (7) 10 (9) 6 (7) 4 (3) 0 (0)
DIP joints (8) 40/17 28/28 48/11 3 (5) 0 (0) 3 (6) 1 (4) 0 (0) 0 (0)
First sternocostal joints (2) 71/0 66/0 77/0 3 (15) 2 (7) 1 (6) 2 (7) 1 (5) 0 (0)
Hips (2) 99/0 97/0 100/0 6 (17) 6 (17) 1 (3) 1 (3) 0 (0) 2 (6)
Knees (2) 100/0 94/0 100/0 15 (42) 6 (17) 4 (11) 3 (10) 2 (6) 0 (0)
Ankles and TMT joints (4) 95/1 91/1 95/1 28 (42) 22 (32) 11 (17) 7 (11) 0 (0) 2 (3)
Feet MTP joints (10) 83/3 73/6 88/1 41 (28) 52 (34) 19 (16) 16 (12) 13 (8) 5 (3)
Feet PIP joints (10) 61/10 32/37 78/3 15 (14) 10 (8) 13 (25) 14 (22) 5 (4) 2 (1)
Feet DIP joints (8) 45/16 10/53 61/10 0 (0) 4 (6) 1 (6) 0 (0) 0 (0) 0 (0)

Spine and SI joints

Fat infiltration BMO Bone erosion

Fat infiltration BMO Bone erosion
All PsA SpA PsA SpA PsA SpA

Cervical vertebral bodies (6) 96/0 93/0 96/0 10 (10) 18 (17) 20 (20) 17 (17) 1 (1) 0 (0)
Cervical posterior parts (6) 96/0 94/0 97/0 1 (1) 0 (0) 5 (5) 0 (0) 0 (0) 0 (0)
Thoracic vertebral bodies (12) 100/0 100/0 100/0 13 (6) 21 (10) 9 (4) 17 (8) 0 (0) 0 (0)
Thoracic posterior parts (12) 100/0 100/0 100/0 0 (0) 1 (0) 3 (1) 2 (1) 0 (0) 0 (0)
Lumbar vertebral bodies (5) 100/0 95/0 100/0 5 (6) 11 (12) 3 (4) 7 (8) 0 (0) 0 (0)
Lumbar posterior parts (5) 100/0 97/0 100/0 0 (0) 0 (0) 1 (1) 0 (0) 0 (0) 0 (0)
SI joint quadrants (8) 94/0 92/1 94/0 17 (12) 44 (31) 5 (4) 30 (21) 14 (10) 30 (21)

a
Concerning readability: areas are divided into readable (left value), in FOV, but not readable (right value), and not in FOV
(value not shown). For example: 79% of MCP joints could be evaluated for BMO, 6% of MCP joints were in FOV but could not
be evaluated for BMO and 15% of MCP joints were not in FOV. bFrequency of lesions is shown as number of detected
lesions, with the frequency (%) per readable area in brackets. A total of 76 peripheral and axial joints were assessed for
synovitis, BMO and bone erosions, and the spine and SI joint were assessed for fat infiltration, BMO and bone erosions.
Spinal DVUs were assessed separately for vertebral bodies and posterior elements. Parentheses in the left column contain
numbers of joints, DVUs or SI joint quadrants assessed. PIP joints include the interphalangeal joints. AC: acromioclavicular;
BMO: bone marrow oedema; DVU: discovertebral unit; FOV: field of view; TMT: tarsometatarsal.

www.rheumatology.oxfordjournals.org 1043
René Panduro Poggenborg et al.

FIG. 1 Examples of whole-body and conventional MRI in SpA and PsA

Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019

(A–F) A 34-year-old male SpA patient with back and knee pain; (G–K) a 32-year-old female with recent onset PsA with
swelling of finger joints. (A) Whole-body MRI (WBMRI) T1-weighted (T1w) fat-saturated (FS) coronal image. (B) WBMRI
post-contrast T1w FS coronal image of right knee, with synovitis at arrow. (C) WBMRI post-contrast T1w FS coronal
image of left knee with synovitis at arrow (zoomed view of left knee shown in (A). (D and E) WBMRI pre- (D) and post- (E)
contrast T1w coronal images of right acromioclavicular (AC) joint (black arrow in D shows AC joint; white arrow in E shows
synovitis). (F) WBMRI T1w coronal image of the SI joints; arrow shows ankylosis. (G) WBMRI T1w coronal image of SI
joints (normal). (H and I) conventional MRI pre- (H) and post- (I) contrast coronal images of second (right) and third (left)
MCP (lower) and PIP (upper) joints, with synovitis at arrows; (J and K) WBMRI pre- (J) and post- (K) contrast coronal
images of the second and third MCP joints shown in (H) and (I), with synovitis at arrows (PIP joints not in field of view).

1044 www.rheumatology.oxfordjournals.org

FIG. 2 Distribution of whole-body MRI inflammatory and structural lesions in PsA (upper row) and SpA (lower row)
For peripheral and axial joints, the frequencies of the findings (in percentage of readable areas) are provided. Results are
summarized for left and right side, with colours indicating frequencies.
in upper quadrants (89 vs 67 lesions), whereas no differ-
ence was found between sacral and iliac sides (76 vs 80
lesions).
In the SI joints, WBMRI detected BMO in two HSs
(Supplementary Fig. S1, available at Rheumatology
Online). convMRI of the SI joint in these two HSs revealed
BMO in one (Berlin score 2 in one quadrant), but no BMO
in the other. Furthermore, convMRI registered BMO (SI
joint Berlin score 1 in two quadrants) in one HS without
BMO by WBMRI. In three HSs, convMRI as well as
WBMRI detected BMO in the cervical vertebrae. They
were a 49-year-old man with a family history of psoriasis,
a 55-year-old woman who had previously had knee arth-
roscopy due to knee pain, and a 61-year-old woman with
a history of surgery for cervical disc herniation.
WBMRI and convMRI scores
WBMRI and convMRI scores, including global scores, for
PsA, SpA and HSs are shown in Table 4. In PsA, WBMRI
showed higher BMO scores than in HSs, while there were
no overall significant differences in other parameters. In
SpA, the global BMO and global structural damage scores
were higher than in HSs. Corresponding convMRI re-
vealed higher synovitis scores in the hands in PsA
www.rheumatology.oxfordjournals.org
Whole-body MRI in PsA and SpA
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
patients, and higher spinal and SI joint structural lesion
scores in SpA patients than in HSs (Table 4).
WBMRI BMO scores for 76 peripheral joints correlated
significantly with WBMRI synovitis (r = 0.48; P = 0.0006)
and erosion (0.51; P = 0.0002). In the spine, WBMRI
BMO scores correlated with fat infiltration (0.45;
P = 0.001). For the WBMRI scores of the SI joints, BMO
correlated with fat infiltration (0.40; P = 0.005) and erosion
(0.51; P = 0.0002), and fat infiltration correlated with ero-
sion (0.41; P = 0.004) and ankylosis (0.33; P = 0.02). The
WBMRI global BMO score correlated significantly with
the WBMRI global structural damage score (r = 0.52;
P = 0.0002).
WBMRI BMO scores for 76 peripheral joints and
WBMRI global BMO correlated with convMRI PsAMRIS
synovitis scores (r = 0.47–0.52; P < 0.05), but not with
BMO scores.
WBMRI and convMRI scores of SI joint BMO, fat infil-
tration and erosion correlated significantly (r = 0.78/0.71/
0.72, all P < 0.0005). WBMRI and convMRI SI joint anky-
losis scores did not correlate significantly (0.24; P = 0.22).
Spine scores of activity and damage by convMRI (Berlin
activity score and ASspiMRI-c scores) and corresponding
WBMRI scores showed lower correlations (r = 0.20–0.55;
0.005 < P < 0.20). Correlation between WBMRI and
1045
René Panduro Poggenborg et al.

TABLE 4 Median (IQR) scores of whole-body MRI and conventional MRI findings in joints, discovertebral units and SI
joint quadrants

Whole-body MRI

PsA SpA Healthy subjects

76-joints: BMO (0–76) 6 (1–9.5)* 5 (0–10) 2 (0–3)

76-joints: synovitis (0–76) 12 (7–14) 10 (3–17) 10.5 (5–18.5)
76-joints: bone erosion (0–76) 0 (0–2) 0 (0–2) 0 (0–1)
Spine: BMO (0–46) 1.5 (0–4) 1 (0–4) 0.5 (0–1.5)
Spine: fat infiltration (0–46) 0 (0–3) 1 (0–4) 0 (0–1)

Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019

Spine: bone erosion (0–46) 0 (0–0) 0 (0–0) 0 (0–0)
SI joint: BMO (0–8) 0 (0–0) 0 (0–3) 0 (0–0)
SI joint: fat infiltration (0–8) 0 (0–0) 0 (0–4) 0 (0–0)
SI joint: bone erosion (0–8) 0 (0–0) 0 (0–2) 0 (0–0)
SI joint: ankylosis (0–8) 0 (0–0) 0 (0–0) 0 (0–0)
Global BMO (0–130) 7 (3–15)* 8 (2–14)* 2.5 (1–4.5)
Global inflammation (0–206) 20.5 (14–27.5) 19.5 (13–31) 11 (8.5–24)
Global structural damage (0–192) 4 (0–6) 7 (3–12)* 1.5 (0–4.5)

Conventional MRI

Hand: PsAMRIS BMO (0–72) 0 (0–0) – 0 (0–0)

Hand: PsAMRIS synovitis (0–36) 2 (1–5)* – 0 (0–0.5)
Hand: PsAMRIS bone erosion (0–120) 0 (0–1) – 0 (0–0)
Spine: Berlin activity (0–69) – 0.5 (0–4) 0 (0–2)
Spine: ASspiMRI-c (0–138) – 5 (1–15) * 0 (0–4)
SI joint: Berlin activity (0–32) – 0 (0–8) 0 (0–0)
SI joint: Berlin structural lesion (0–8) – 2.5 (0–6)* 0 (0–0)

*P < 0.05, Mann–Whitney test vs healthy subjects. Possible ranges of scores are shown in parentheses after each feature.
Global inflammation score is the sum of synovitis and global BMO score. Global structural damage score is the sum of bone
erosions, fat infiltration and ankylosis. WBMRIs with very low readability (<33%) for assessed features were excluded. 76-
joints: joints included in the 78-tender joint count except for the TM joints; ASspiMRI-c: AS spine MRI scoring system
chronicity score; BMO: bone marrow oedema; DVU: discovertebral unit; PsAMRIS: PsA MRI scoring system.

convMRI scores of synovitis in hands did not reach stat-
istical significance (0.32; P = 0.15).
Association between MRI and clinical examination
WBMRI and convMRI findings were compared with clin-
ical examination. The 28-SJC correlated significantly with
BMO assessed by WBMRI in the same 28 joints (all par-
ticipants: r = 0.31; P = 0.04; PsA-only: r = 0.54; P = 0.03).
In PsA, the SJC and scores of BMO in 76 peripheral joints
correlated significantly (r = 0.66; P = 0.006). WBMRI syno-
vitis was not significantly correlated with SJC or TJC.
Apart from weak correlations between WBMRI BMO and
erosion scores for 76 peripheral joints with age (both 0.30,
P < 0.05), no significant correlations between peripheral
WBMRI scores and patient characteristics or clinical
measures of disease activity were found. The WBMRI
scores of spinal fat infiltration correlated with BASMI
(0.35, P = 0.02), and the SI joint damage score with
BASFI (0.29, P = 0.049), whereas no other significant cor-
relations were observed.
The WBMRI global structural damage correlated with
BASMI (0.37, P = 0.016), whereas no other significant cor-
relations were observed with patient characteristics,
including clinical measures of disease activity.
By convMRI, the PsAMRIS synovitis score correlated
statistically significantly with 76-SJC (r = 0.45; P = 0.04).
The ASspiMRI-c score assessed by convMRI of the
spine correlated with BASDAI, BASMI and BASFI
(r = 0.42; P = 0.023/r = 0.50; P = 0.0086/r = 0.57;
P = 0.0016). No further correlations were found between
convMRI scores in the spine and SI joint and clinical
outcomes.
WBMRI reliability analysis
The intrareader reproducibility of WBMRI assessments
was assessed, comparing original scores with re-
anonymized re-scorings of 10 study participants 12
months later. Intrareader ICCs were 0.31–0.85 for syno-
vitis, BMO and erosion in the 76 peripheral joints,
0.62–0.68 for BMO and fat infiltration in the spine, and
0.81–1.0 for BMO, fat infiltration, erosion and ankylosis
in the SI joints.
Discussion
This prospective pilot study of HSs and patients with PsA
or axial SpA demonstrated the ability of head-to-toe
WBMRI to assess overall disease activity and structural
damage. The readability was good in axial and proximal

1046 www.rheumatology.oxfordjournals.org

peripheral joints, while it decreased in more distally
located peripheral joints. The pattern of inflammatory
and structural changes in SpA and PsA was described.
SI joint involvement was most frequent in SpA, while per-
ipheral joint involvement, particularly in knees and hand
MCP and PIP joints, was most frequent in PsA. WBMRI
findings correlated moderately with clinical findings. The
reproducibility of the measurements was dependent on
the measured parameter and site, with the best result
for axial joints.
A detailed evaluation of the ability of WBMRI to visualize
inflammatory and structural changes in different parts of
the body was undertaken. Some areas, such as hips,
knees, spine and SI joints, were almost always success-
fully visualized, whereas other areas were more diffi-
cult—including the elbows and distal joints of the hands
and feet. Poorer readability had several causes, including
movement artefacts, longer distance from the centre of
the MRI unit decreasing field homogeneity (off-centre
artefact) and consequently image quality, and smaller
size of the joint causing insufficient spatial resolution.
Areas like the spine, SI joint and knee were favoured by
being positioned close to the coil and the centre of the
magnetic field, providing very good readability. Although
image acquisition was not consistently of sufficient quality
for robust and reliable assessment of all joints of interest,
the study demonstrated important aspects of joint in-
volvement in PsA and SpA and illustrated the large
future potential of WBMRI to assess these anatomically
widespread diseases. Improved hardware becoming
available currently and in the future will undoubtedly in-
crease the clinical utility of WBMRI, as will further refined
techniques of image acquisition and evaluation.
Furthermore, more optimal positioning of the patient in
the scanner may facilitate reliable assessment of periph-
eral joints.
We described the pattern of WBMRI joint involvement in
PsA and SpA (Fig. 2). Only a few studies have investigated
head-to-toe WBMRI in prospective studies of patients
with PsA [7] or SpA [8–10]. In agreement with these, we
found widespread inflammation. Inflammatory lesions out-
side the axial joints were frequent, in both SpA and PsA,
which is in agreement with a previous WBMRI pilot study
of SpA patients [21]. In PsA, the joints most frequently
involved were wrists, shoulders, ankles, knees, MCP
and MTP joints. The same pattern was seen in SpA,
except that MCP joints were replaced with hip joints.
Spine and SI joint involvement were seen in both dis-
eases, but most frequently in SpA, as expected when
the PsA population was selected based on peripheral in-
volvement. The DIP joints are often clinically affected in
PsA [22], but we rarely detected DIP synovitis or BMO.
The fact that some joints were involved to a different
extent than expected, for example, that MCP involvement
was frequent while DIP joint involvement was rare, may be
related both to WBMRI revealing subclinical disease and
to the fact that very small joints are more difficult to assess
by WBMRI, and therefore may be falsely negative. Future
studies, with larger sample sizes and improved
www.rheumatology.oxfordjournals.org
Whole-body MRI in PsA and SpA
techniques, are needed to confirm the pattern of joint
involvement.
In the present study, BMO was registered by WBMRI in
some HSs. This indicates the need to define appropriate
definitions and cut-offs of pathology on WBMRI in future
studies before the method is used in clinical practice.
Nevertheless, BMO was markedly more frequent in PsA
and SpA. Of particular importance are the findings of BMO
in SI joints, as such changes are part of the definition of
axial SpA in the ASAS criteria for axial SpA [23, 24]. Both
convMRI and WBMRI detected BMO in two SI joints of
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
HSs. The extent was, however, limited to 1–2 quadrants.
These findings are in agreement with previous convMRI
studies in HSs, which report that BMO can be seen in up
to one-quarter of healthy controls [25].
The correlation between findings by MRI and by clinical
examination was investigated. BMO scores correlated
significantly with SJC in PsA, supporting the construct
validity of WBMRI BMO scores. In contrast, no associ-
ation was found between WBMRI synovitis and clinical
joint counts. Furthermore, the intrareader reproducibility
of peripheral synovitis was low. This indicates that
WBMRI assessment of synovitis was difficult in several
joints, even though the results probably also partly reflect
detection of subclinical inflammation. convMRI is an es-
tablished method for assessing joint inflammation, and the
WBMRI technique differs from convMRI mainly in the
much larger area visualized, at the cost of lower image
resolution. Probably, higher image resolution and quality
than available in the present study will markedly improve
the WBMRI assessment of synovitis.
We found a correlation between the SI joint structural
damage score and BASFI. We also found that WBMRI
spinal BMO and spinal fat infiltration correlated with the
metrology index BASMI. This finding adds to an earlier
investigation by Althoff et al. [26], who found a correlation
between WBMRI inflammation and clinical symptoms. In
contrast, Weber et al. [27] found no association between
WBMRI assessments and clinical outcomes. In a recent
study of WBMRI during anti-TNF treatment, it was re-
ported that change in BASDAI and BASFI correlated
with changes in MRI SI joint scores, whereas no correl-
ation was found between MRI spine scores and clinical
parameters [9].
We found a moderate agreement between WBMRI and
convMRI scores. In SpA, investigation of inflammatory SI
joint lesions by WBMRI has previously been compared
with convMRI, finding a strong correlation and compar-
able reliability between the two methods, but no correl-
ation with clinical measures [28]. This study, however, did
not examine head-to-toe, but only the SI joints and spine,
providing an overall higher imaging quality. Furthermore, it
should be noted that the true gold standard for WBMRI is
not convMRI, but rather biopsies for inflammatory fea-
tures, and CT for structural lesions [29]. Future studies
including such standard reference methods would be
highly relevant.
In the present study, structural lesion scores by
convMRI (ASspiMRI-c score and Berlin structural lesion
1047
René Panduro Poggenborg et al.
SI joint score) were significantly higher in SpA than in HSs
(as were the WBMRI global structural damage scores),
and they were also correlated with BASMI, BASFI and
BASDAI. Overall, the present and earlier studies have
found a low to moderate correlation with clinical findings,
indicating that WBMRI does more than provide informa-
tion already available.
The reliability of WBMRI assessments ranged from poor
to excellent for different features. The limited reproduci-
bility in some areas is a consequence of acquiring images
at the limits of what is possible technically, and perfect
visualization of all features by WBMRI was not expected
prior to study initiation. Re-reading of the WBMRIs was
performed 1 year after the original readings, and as
WBMRI assessment is not routine work, the internal cali-
bration of the reader may have declined between read-
ings. Only a few studies have reported ICCs for WBMRI
[9, 10, 30]. However, although high ICCs (>0.85) were re-
ported therein, the reliability in those studies was tested
using different techniques. In particular, dedicated coils
were applied, positioned close to the various imaged
parts of the patients, while such coils were not available
in our scanner (in which images were acquired using the
integrated quadrature body coil of the unit). Furthermore,
the reliabilities reported did not reflect assessment of the
entire body but only of the spine, SI joints and, in one case
[10], pelvic entheses. To optimize reliability, improved
imaging acquisition, revised precise definitions of the
scored features, and more extensive reader training and
calibration are needed.
In conclusion, this study showed that WBMRI allows
assessment of peripheral and axial joints in patients with
PsA and SpA. Readability was high for spine, SI joints and
proximal peripheral joints, but lower for more distally
located joints. Patterns of joint involvement were identi-
fied. WBMRI is a very promising tool for global assess-
ment of disease activity and damage in PsA and SpA, and
the present study encourages further research aiming at
optimization of image acquisition, and assessment and
validation in longitudinal studies.
Acknowledgements
The authors acknowledge the assistance of study partici-
pants, radiographers, study nurses and laboratory staff
who participated in the study. The study was conducted
without any financial support. René Panduro Poggenborg
has during his Ph.D. study received grants from The
Danish Rheumatism Association, the Scandinavian
Journal of Rheumatology and The Danish Psoriasis
Research Foundation, and was furthermore supported
by The Capital Region of Denmark.
Disclosure statement: The authors have declared no
conflicts of interest.
Supplementary data
Supplementary data are available at Rheumatology
Online.
1048
References
1 Østergaard M, Poggenborg RP, Axelsen MB et al.
Magnetic resonance imaging in spondyloarthritis—how to
quantify findings and measure response. Best Pract Res
Clin Rheumatol 2010;24:637–57.
2 Dougados M, Baeten D. Spondyloarthritis. Lancet 2011;
377:2127–37.
3 Mease PJ. Psoriatic arthritis: update on pathophysiology,
assessment and management. Ann Rheum Dis 2011;
70(Suppl 1):i77–84.
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
4 Meaney JF, Fagan A. Whole-body MR imaging in a
multimodality world: current applications, limitations,
and future potential for comprehensive musculo-
skeletal imaging. Semin Musculoskelet Radiol 2010;14:
14–21.
5 Mager AK, Althoff CE, Sieper J et al. Role of whole-body
magnetic resonance imaging in diagnosing early spondy-
loarthritis. Eur J Radiol 2009;71:182–8.
6 Poggenborg RP, Eshed I, Ostergaard M et al. Enthesitis in
patients with psoriatic arthritis, axial spondyloarthritis and
healthy subjects assessed by ‘head-to-toe’ whole-body
MRI and clinical examination. Ann Rheum Dis 2014,
Advance access published 3 January 2014,
doi: 10.1136/annrheumdis-2013-204239.
7 Weckbach S, Schewe S, Michaely HJ et al. Whole-body
MR imaging in psoriatic arthritis: additional value for
therapeutic decision making. Eur J Radiol 2011;77:
149–55.
8 Althoff CE, Sieper J, Song IH et al. Active inflammation
and structural change in early active axial spondyloarthritis
as detected by whole-body MRI. Ann Rheum Dis 2013;72:
967–73.
9 Song IH, Hermann K, Haibel H et al. Effects of etanercept
versus sulfasalazine in early axial spondyloarthritis on
active inflammatory lesions as detected by whole-body
MRI (ESTHER): a 48-week randomised controlled trial.
Ann Rheum Dis 2011;70:590–6.
10 Karpitschka M, Godau-Kellner P, Kellner H et al.
Assessment of therapeutic response in ankylosing
spondylitis patients undergoing anti-tumour necrosis
factor therapy by whole-body magnetic resonance
imaging. Eur Radiol 2013;23:1773–84.
11 Moll JM, Wright V. Psoriatic arthritis. Semin Arthritis
Rheum 1973;3:55–78.
12 Dougados M, van der Linden S, Juhlin R et al. The
European Spondylarthropathy Study Group preliminary
criteria for the classification of spondylarthropathy.
Arthritis Rheum 1991;34:1218–27.
13 Garrett S, Jenkinson T, Kennedy LG et al. A new approach
to defining disease status in ankylosing spondylitis: the
Bath Ankylosing Spondylitis Disease Activity Index.
J Rheumatol 1994;21:2286–91.
14 Pedersen OB, Hansen GO, Svendsen AJ et al. Adaptation
of the Bath measures on disease activity and function in
ankylosing spondylitis into Danish. Scand J Rheumatol
2007;36:22–7.
15 Gladman DD, Helliwell P, Mease PJ et al. Assessment of
patients with psoriatic arthritis: a review of currently
available measures. Arthritis Rheum 2004;50:24–35.
www.rheumatology.oxfordjournals.org

16 Boyesen P, McQueen FM, Gandjbakhch F et al. The
OMERACT Psoriatic Arthritis Magnetic Resonance
Imaging Score (PsAMRIS) is reliable and sensitive to
change: results from an OMERACT workshop.
J Rheumatol 2011;38:2034–8.
17 Ostergaard M, McQueen F, Wiell C et al. The OMERACT
psoriatic arthritis magnetic resonance imaging scoring
system (PsAMRIS): definitions of key pathologies, sug-
gested MRI sequences, and preliminary scoring system
for PsA Hands. J Rheumatol 2009;36:1816–24.
18 Haibel H, Rudwaleit M, Brandt HC et al. Adalimumab re-
duces 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.
19 Braun J, Baraliakos X, Golder W et al. Magnetic resonance
imaging examinations of the spine in patients with anky-
losing spondylitis before and after successful therapy with
infliximab: evaluation of a new scoring system. Arthritis
Rheum 2003;48:1126–36.
20 Hermann KG, Braun J, Fischer T et al. [Magnetic reson-
ance tomography of sacroiliitis: anatomy, histological
pathology, MR-morphology, and grading]. Radiologe
2004;44:217–28.
21 Weber U, Pfirrmann CW, Kissling RO et al. Whole body
MR imaging in ankylosing spondylitis: a descriptive pilot
study in patients with suspected early and active con-
firmed ankylosing spondylitis. BMC Musculoskelet Disord
2007;8:20.
22 Gladman DD, Antoni C, Mease P, Clegg D, Nash P.
Psoriatic arthritis: epidemiology, clinical features, course,
and outcome. Ann Rheum Dis 2005;64(Suppl 2):ii14–17.
23 Rudwaleit M, van der Heijde D, Landewe R et al. The
development of Assessment of SpondyloArthritis
www.rheumatology.oxfordjournals.org
Whole-body MRI in PsA and SpA
international Society classification criteria for axial spon-
dyloarthritis (part II): validation and final selection. Ann
Rheum Dis 2009;68:777–83.
24 Rudwaleit M, Jurik AG, Hermann KG et al. Defining active
sacroiliitis on magnetic resonance imaging (MRI) for clas-
sification of axial spondyloarthritis: a consensual approach
by the ASAS/OMERACT MRI group. Ann Rheum Dis 2009;
68:1520–7.
25 Weber U, Maksymowych WP. Sensitivity and specificity of
magnetic resonance imaging for axial spondyloarthritis.
Am J Med Sci 2011;341:272–7.
Downloaded from https://academic.oup.com/rheumatology/article-abstract/54/6/1039/1799654 by guest on 15 March 2019
26 Althoff CE, Appel H, Rudwaleit M et al. Whole-body MRI
as a new screening tool for detecting axial and peripheral
manifestations of spondyloarthritis. Ann Rheum Dis 2007;
66:983–5.
27 Weber U, Lambert RG, Rufibach K et al. Anterior chest
wall inflammation by whole-body magnetic resonance
imaging in patients with spondyloarthritis: lack of associ-
ation between clinical and imaging findings in a cross-
sectional study. Arthritis Res Ther 2012;14:R3.
28 Weber U, Maksymowych WP, Jurik AG et al. Validation of
whole-body against conventional magnetic resonance
imaging for scoring acute inflammatory lesions in the
sacroiliac joints of patients with spondylarthritis. Arthritis
Rheum 2009;61:893–9.
29 Østergaard M, Wiell C. Ultrasonography in rheumatoid
arthritis: a very promising method still needing more val-
idation. Curr Opin Rheumatol 2004;16:223–30.
30 Song IH, Hermann KG, Haibel H et al. Relationship
between active inflammatory lesions in the spine and
sacroiliac joints and new development of chronic lesions
on whole-body MRI in early axial spondyloarthritis: results
of the ESTHER trial at week 48. Ann Rheum Dis 2011;70:
1257–63.
1049