doi:10.1093/brain/aww037
BRAIN 2016: 139; 1472–1481
| 1472
Increased cortical grey matter lesion
detection in multiple sclerosis with 7 T MRI:
a post-mortem verification study
Iris D. Kilsdonk,1, Laura E. Jonkman,2, Roel Klaver,2 Susanne J. van Veluw,3
Jaco J. M. Zwanenburg,4 Joost P. A. Kuijer,5 Petra J. W. Pouwels,5 Jos W. R. Twisk,6
Mike P. Wattjes,1 Peter R. Luijten,3 Frederik Barkhof1 and Jeroen J. G. Geurts2
These authors contributed equally to this work.
The relevance of cortical grey matter pathology in multiple sclerosis has become increasingly recognized over the past decade.
Unfortunately, a large part of cortical lesions remain undetected on magnetic resonance imaging using standard field strength. In
vivo studies have shown improved detection by using higher magnetic field strengths up to 7 T. So far, a systematic histopathological verification of ultra-high field magnetic resonance imaging pulse sequences has been lacking. The aim of this study was to
determine the sensitivity of 7 T versus 3 T magnetic resonance imaging pulse sequences for the detection of cortical multiple
sclerosis lesions by directly comparing them to histopathology. We obtained hemispheric coronally cut brain sections of 19 patients
with multiple sclerosis and four control subjects after rapid autopsy and formalin fixation, and scanned them using 3 T and 7 T
magnetic resonance imaging systems. Pulse sequences included T1-weighted, T2-weighted, fluid attenuated inversion recovery,
double inversion recovery and T2 . Cortical lesions (type I–IV) were scored on all sequences by an experienced rater blinded to
histopathology and clinical data. Staining was performed with antibodies against proteolipid protein and scored by a second reader
blinded to magnetic resonance imaging and clinical data. Subsequently, magnetic resonance imaging images were matched to
histopathology and sensitivity of pulse sequences was calculated. Additionally, a second unblinded (retrospective) scoring of
magnetic resonance images was performed. Regardless of pulse sequence, 7 T magnetic resonance imaging detected more cortical
lesions than 3 T. Fluid attenuated inversion recovery (7 T) detected 225% more cortical lesions than 3 T fluid attenuated inversion
recovery (Z = 2.22, P 5 0.05) and 7 T T2 detected 200% more cortical lesions than 3 T T2 (Z = 2.05, P 5 0.05). Sensitivity of 7
T magnetic resonance imaging was influenced by cortical lesion type: 100% for type I (T2), 11% for type II (FLAIR/T2), 32% for
type III (T2 ), and 68% for type IV (T2). We conclude that ultra-high field 7 T magnetic resonance imaging more than doubles
detection of cortical multiple sclerosis lesions, compared to 3 T magnetic resonance imaging. Unfortunately, (subpial) cortical
pathology remains more extensive than 7 T magnetic resonance imaging can reveal.
1
2
3
4
5
6
Department
Department
Department
Department
Department
Department
of
of
of
of
of
of
Radiology and Nuclear Medicine, VU University Medical Centre, Amsterdam, The Netherlands
Anatomy and Neurosciences, VU University Medical Centre, Amsterdam, The Netherlands
Neurology, University Medical Centre Utrecht, Utrecht, The Netherlands
Radiology, University Medical Centre Utrecht, Utrecht, The Netherlands
Physics and Medical Technology, VU University Medical Centre, Amsterdam, The Netherlands
Epidemiology and Biostatistics, VU University Medical Centre, Amsterdam, The Netherlands
Correspondence to: Iris D. Kilsdonk,
Department of Radiology and Nuclear Medicine,
VU Medical Centre Amsterdam,
PO Box 7075, 1007 MB Amsterdam, The Netherlands
Received August 20, 2015. Revised December 14, 2015. Accepted January 20, 2016. Advance Access publication March 8, 2016
ß The Author (2016). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.
For Permissions, please email: [email protected]
Post-mortem MS cortical lesion detection with 7 T MRI
BRAIN 2016: 139; 1472–1481
| 1473
E-mail: [email protected]
Keywords: multiple sclerosis; 7 Tesla MRI; cortical lesions; subpial demyelination; ultrahigh field MRI; histopathology
Abbreviations: DIR = double inversion recovery; FLAIR = fluid attenuated inversion recovery
Introduction
In the past decade, the focus in multiple sclerosis research
has switched from white matter to grey matter involvement
(Vigeveno et al., 2012). We now know that cortical grey
matter lesions are common (Kidd et al., 1999; Geurts et al.,
2005; Kutzelnigg et al., 2005) and are related to physical
and cognitive disability in multiple sclerosis patients
(Kutzelnigg and Lassmann, 2006; Roosendaal et al.,
2009). Cortical pathology can be present early in the disease (Calabrese and Gallo, 2009; Lucchinetti et al., 2011),
accumulates with disease duration and is most prominent
in the progressive phase (Kutzelnigg et al., 2005). The detection of cortical lesions could be highly relevant in the
clinical setting, as it may be used for (differential) diagnostic and disease monitoring purposes.
Unfortunately, a large part of cortical lesions go undetected on conventional MRI using standard field strength
(Daams et al., 2013). The most likely reason for the difficulty of detecting cortical lesions is their small size
(Seewann et al., 2011). Also the difference in pathological
substrate (relative lack of inflammation compared to white
matter lesions), anatomical paucity of myelin in the cortex
generating little MRI contrast upon demyelination, and
partial volume effects from adjacent CSF and white
matter probably play a role (Kidd et al., 1999; Peterson
et al., 2001). Improvements have been made by developing
grey matter specific pulse sequences such as double inversion recovery (DIR) or phase sensitive inversion recovery
(Geurts et al., 2005; Wattjes et al., 2007; Simon et al.,
2010; Seewann et al., 2012; Sethi et al., 2012), and by
moving to high field 3 T and ultra-high field 7 T MRI
systems (Wattjes and Barkhof, 2009; Kilsdonk et al.,
2012; Filippi et al., 2014).
The implementation of 7 T MRI resulted in an
increased detection of cortical lesions in multiple sclerosis
patients, compared to lower 3 T and 1.5 T MRI systems
(Kollia et al., 2009; Mainero et al., 2009; Tallantyre et
al., 2010; De Graaf et al., 2013), mostly due to superior
signal-to-noise ratios, spatial resolution and image contrast. Though these first results at 7 T seemed promising,
the question still remains what proportion of the actual
number of (histologically observed) cortical lesions were
picked up.
The aim of the present study was to determine and
compare the sensitivities of a multi-contrast MRI protocol
at 3 T and 7 T for the detection of cortical lesions in
multiple sclerosis, by directly comparing magnetic resonance images to histopathology in a unique post-mortem
dataset.
Materials and methods
Patients and autopsy procedure
Coronally cut, 10-mm thick full-hemispheric brain slices of 19
multiple sclerosis patients were formalin-fixed after rapid autopsy [mean post-mortem delay 5 h 56 min, standard deviation
(SD) 2 h 27 min]. Prior to death, multiple sclerosis patients had
been registered at the Netherlands Brain Bank, Amsterdam,
The Netherlands. Additionally, four control slices without
macroscopically visible pathology were obtained from donors
recruited by the Department of Pathology of the VU University
Medical Centre (Amsterdam, The Netherlands). Controls were
not part of the rapid autopsy procedure, hence a longer postmortem delay is present in this group. Table 1 provides demographic details of the donors. All donors gave written informed consent for the use of their tissue and medical
records for research purposes. Permission for performing
autopsies, use of tissue and access to medical records was
granted by the institutional ethics review board.
MRI
A custom-built perspex slice holder was used to scan three
slices simultaneously, immersed in 10% formalin. Multicontrast MRI was performed using wholebody 3 T (Philips
Achieva) and 7 T systems (Philips Achieva). At 3 T, the standard 8-channel SENSE head coil was used for imaging. At 7 T,
a volume transmit head coil with 32-channel receiver coil
(Nova Medical Inc.) was used. Pulse sequences included were
the clinically applied T1-weighted, T2-weighted and fluid attenuated inversion recovery (FLAIR) (Kilsdonk et al., 2013) plus
pulse sequences that have shown to improve cortical lesion
detection at 3 T and at 7 T in a research setting, i.e. DIR
and T2 . Examples of the multi-contrast images can be
seen in Fig. 1, and the sequence parameters are displayed in
Table 2. Inversion delays were chosen to null the formalin
(T1-weighted and FLAIR), and both the formalin and the
white matter (DIR).
For the 7 T T1-weighted images, we performed an offline
real-part reconstruction. A B0-map was used to correct for
off-resonance effects while linear phase correction, using two
reference points in the surrounding formalin, was applied to
correct for echo shifting along the readout direction.
Histology
After MRI acquisition, the full-hemispheric brain slices were cut
in half coronally, and were manually dehydrated through a
series of 70%, 96%, 100% alcohol (ethanol) and xylene, and
embedded in paraffin. Sections (8-mm thick) were cut, mounted
onto glass slides (Superfrost, VWR international) and dried
overnight at 37 C. Sections were deparaffinated an rehydrated
in a series of xylene, 100%, 96% and 70% ethanol and rinsed
1474
| BRAIN 2016: 139; 1472–1481
I. D. Kilsdonk et al.
Table 1 Demographics of patients with multiple sclerosis and control subjects
Case
Sex
Multiple sclerosis
1
M
2
F
3
M
4
F
5
M
6
F
7
M
8
M
9
M
10
M
11
F
12
F
13
F
14
M
15
F
16
F
17
F
18
M
19
F
Mean (SD)
Control
20
F
21
F
22
F
23
F
Mean (SD)
Age (years)
80
81
75
66
71
54
63
78
59
56
56
66
54
58
95
81
85
60
67
68 ( 12)
72
58
76
76
71 ( 9)
PMD (h:min)
6:05
3:30
6:10
7:30
4:00
6:00
4:30
3:00
5:00
6:10
8:25
9:35
3:30
6:00
6:30
6:30
5:00
2:00
7:30
5:56 ( 2:27)
424:00
524:00
524:00
58:00
DD (years)
45
27
50
17
15
16
25
33
Unknown
13
32
23
31
27
55
21
58
31
9
32 ( 16)
NA
NA
NA
NA
Multiple sclerosis disease type
Cause of death
SPMS
PPMS
Unknown
Unknown
SPMS
SPMS
SPMS
SPMS
SPMS
Unknown
SPMS
PPMS
SPMS
SPMS
SPMS
SPMS
Unknown
SPMS
Unknown
Pneumonia
Pneumonia
Pneumonia
Pulmonary hypertension
Pulmonary cancer
Cholangiocarcinoma
Unknown
Euthanasia
Euthanasia
Suicide (medication)
Pneumonia
Euthanasia
Heart failure
Pneumonia
Unknown
Heart failure
Stroke
Pneumonia
Sepsis
Stroke
Breast cancer
Pneumonia
Pneumonia
DD = disease duration; F = female; M = male; NA = not applicable; PMD = post-mortem delay; PPMS = primary progressive multiple sclerosis; SPMS = secondary progressive multiple sclerosis.
Figure 1 Images of a single brain slice of a multiple sclerosis patient obtained with pulse sequences at 7 T and 3 T. T2 = T2 weighted; T1 = T1-weighted; T2 = T2-weighted.
with 0.05 M TBS (tris-buffered saline pH 7.8–8.0). Endogenous
peroxidase activity was blocked by incubating the sections in
TBS with 0.3% H2O2 for 30 min. After this, the sections were
rinsed with 0.1 M PBS (phosphate-buffered saline, pH 7.4).
Staining was performed with antibodies against PLP (proteolipid protein; Serotec) diluted in TBS (1:500) and incubated
overnight at 4 C. After washing, bound primary antibodies
were detected using EnVisionÕ (DAKOCytomation) for 30 min
3500
164
0.40 0.40
0.40
0.39 0.39
0.20
1:52
1500
146
0.65 0.65
0.65
0.50 0.50
0.65
0:22
7.7
3.5
280
0.40 0.40
0.40
0.40 0.40
0.40
0:55
| 1475
Scoring, classification and matching
75
20
0.18 0.18
0.18
0.17 0.17
0.17
4:59
41.8
59
0.32 0.32
0.32
0.30 0.30
0.32
0:39
8000
164
1600
0.40 0.40
0.40
0.39 0.39
0.20
4:16
4000
146
800
0.65 0.65
0.65
0.50 0.50
0.65
0:30
T2 w = T2 -weighted; T1w = T1-weighted; T2w = T2-weighted.
8000
194
2275/325
0.40 0.40
0.40
0.39 0.39
0.20
5:36
4000
146
800/50
0.65 0.65
0.65
0.50 0.50
0.65
0:30
Repetition time (ms)
Echo time (ms)
Inversion time (1/2) (ms)
Acquisition resolution, inplane (mm2)
Acquisition resolution, through plane (mm)
Rec. resolution, inplane (mm2)
Rec. resolution, throughplane (mm)
Acquisition time (h:min)
BRAIN 2016: 139; 1472–1481
at room temperature. Peroxidase activity was demonstrated
with 0.5 mg/ml 3,3’ DAB (diaminobenzidine tetrahydrochloride;
Sigma) in 0.05 M tris-hydrochloride containing 0.03% H2O2
for 5 min, which led to a brown reaction product. Sections were
counterstained with haematoxylin (Sigma) and mounted using
DepexÕ (BDH).
5.2
2.3
300
0.65 0.65
0.65
0.50 0.50
0.65
0:33
7 T
3 T
7T
3 T
3 T
3 T
3T
Sequence
Table 2 Sequence parameters at 3 T and 7 T
7 T
FLAIR
DIR
7 T
T2 w
7 T
T1w
T2 w
Post-mortem MS cortical lesion detection with 7 T MRI
MRI lesions were marked on all pulse sequences separately
(DIR, FLAIR, T2 , T1-weighted and T2-weighted) using
Medical Image Processing, Analysis and Visualization software
(MIPAV, Centre for Information Technology, National
Institutes of Health, Bethesda, MD, USA). The MRI reader
(I.D.K., 4 years’ experience) was blinded to clinical data and
histopathological results. MRI lesions were marked throughout
all the MRI slices, to avoid bias in only scoring within the
histopathologically sampled areas, i.e. before matching MRI
to the histopathological coupes.
Histopathological lesions were defined as areas of complete
demyelination (lack of PLP) and scored by a pathology reader
(L.E.J., 4 years’ experience) who was blinded to the MRI and
clinical data. On MRI as well as histology, cortical lesions
were classified according to criteria introduced by Bø et al.
(2003), in which a distinction between four cortical lesion
types is described; type I, also called leukocortical lesions,
which involve the deeper layers of the grey matter as well
as the adjacent white matter at the grey/white matter junction; type II, or intracortical lesions, which are small demyelinated lesions often centred around blood vessels and
confined within the cortex; type III, extending from the pial
surface into the cortex. When lesions extend to the entire
width of the cortex, they are defined as type IV lesions.
Furthermore, white matter lesions were scored. After MRI
and histopathological scoring, the PLP tissue sections were
carefully matched to the corresponding MRI planes, using
white matter lesions and as many cortical anatomical landmarks as possible. After the blinded, prospective scoring of
the post-mortem MRI and the tissue-to-MRI matching that
was performed in consensus between the two observers,
histopathology scores were made available to the MRI
reader and a second, retrospective, unblinded scoring was
performed (scoring magnetic resonance images with histopathological lesion location and lesion type known to the
readers).
Statistical analysis
Histopathological lesion count was considered the point of reference. Sensitivity of the MRI sequences for detecting lesions
was defined as the percentage of lesions scored in the prospective or retrospective ratings relative to the number of lesions
determined by histopathology. Sensitivity of MRI pulse sequences was statistically compared between and within the
two different field strengths, with a negative binominal multilevel analysis for correlated (non-independent) count data with
a large standard deviation using MLwiN version 2.22 (Rasbash,
2009). P-values of 5 0.05 were considered as statistically
significant.
1476
| BRAIN 2016: 139; 1472–1481
I. D. Kilsdonk et al.
Results
Of the 19 multiple sclerosis and four control coronal hemispheric brain slices, two had to be discarded due to suboptimal matching with MRI, resulting from tissue
processing (lack of adhesion onto the glass slide). This resulted in a final dataset of 17 multiple sclerosis patients and
four controls. Three patients did not show any histopathological abnormalities on the available slice. The control
slices did not show any histopathological and MRI
abnormalities besides changes due to normal ageing. In
the multiple sclerosis brain slices, we identified a total of
99 lesions with histology: 16 white matter lesions and 83
cortical lesions. Of these cortical lesions, six were mixed
grey matter–white matter (type I) lesions and 77 were
located entirely within the cortical grey matter (27 type
II, 28 type III and 22 type IV lesions). Results of histopathological lesion count and the proportion of lesions detected on MRI are shown in Table 3 (prospective scoring)
and Table 4 (retrospective scoring).
Prospective lesion detection
Prospectively, at 7 T, total lesion detection (white matter + grey matter) was 59% higher than at 3 T, with 38
lesions averaged over the pulse sequences versus 24, respectively (Z = 3.41, P 5 0.001). DIR (7 T) detected 24%
more lesions than 3 T DIR (36 versus 29), 7 T FLAIR
detected 100% more lesions than 3 T FLAIR (40 versus
20), 7 T T2 detected 133% more lesions than 3 T T2 (35
versus 15), 7 T T1 detected 30% more lesions than 3 T T1
(35 versus 27) and 7 T T2 detected 52% more lesions than
3 T T2 (41 versus 27) (Table 3). This difference in total
lesion detection was significant for FLAIR (Z = 2.23,
P = 0.026) and T2 (Z = 2.46, P = 0.014). Within field
strength, there was no sequence that detected significantly
more lesions than other sequences.
When focusing on grey matter lesions (I–IV), 7 T FLAIR
detected 225% more lesions than 3 T FLAIR (26 versus 8;
Z = 2.22, P = 0.026) and 7 T T2 detected 200% more grey
matter lesions than 3 T T2 (24 versus 8; Z = 2.05,
P = 0.040). When only focusing on intracortical grey
matter lesions (II–IV), 7 T FLAIR detected 340% more
lesions than 3 T FLAIR (22 versus 5; Z = 2.41,
P = 0.014) and 7 T T2 detected 250% more grey matter
lesions than 3 T T2 (23 versus 11; Z = 2.22, P = 0.026).
With regards to white matter lesion detection, no significant differences were found between 3 T and 7 T for any
sequence.
We identified several MRI abnormalities that were not
confirmed as multiple sclerosis lesions upon histopathology;
52 across all five sequences at 7 T. This comes down to an
average of 3.1 ‘false positives’ per patient (DIR: 0.6;
FLAIR: 0.7; T2 : 0.6; T1: 0.4; T2: 0.8). Additionally, we
identified 48 false positives across all five sequences at 3
T, which is an average of 2.9 per patient (DIR: 0.6; FLAIR:
0.5; T2 : 0.5; T1: 0.5; T2: 0.8). Many of these false positives
were overlapping between sequences (i.e. the same lesion
Table 3 Lesion count (sensitivity in %) prospective MRI scoring
Histology
Prospective rating MRI
Lesion type
n
3 T DIR
7 T DIR
3 T FLAIR
7 T FLAIR
3 T T2
7 T T2
3 T T1
7 T T1
3 T T2
7 T T2
Type I
Type II
Type III
Type IV
I–IV
WML
Total
6
27
28
22
83
16
99
5
1
2
9
17
12
29
5
1
2
14
22
14
36
3
0
0
5
8
12
20
4
3
6
13
26
14
40
2
0
0
6
8
7
15
3
0
9
12
24
11
35
2
0
2
10
14
13
27
4
2
4
14
24
11
35
3
1
3
7
14
13
27
6
3
5
15
29
12
41
(83)
(4)
(7)
(41)
(20)
(75)
(29)
(83)
(4)
(7)
(64)
(27)
(88)
(36)
(50)
(0)
(0)
(23)
(10)
(75)
(20)
(67)
(11)
(21)
(59)
(31)
(88)
(40)
(33)
(0)
(0)
(27)
(10)
(44)
(15)
(50)
(0)
(32)
(55)
(29)
(69)
(35)
(33)
(0)
(7)
(46)
(17)
(81)
(27)
(67)
(7)
(14)
(64)
(29)
(69)
(35)
(50)
(4)
(11)
(32)
(17)
(81)
(27)
(100)
(11)
(18)
(68)
(35)
(75)
(41)
Differences in lesion count between 3 T and 7 T that are statistically significant (P 5 0.05) are in bold font. WML = white matter lesion.
Table 4 Lesion count (sensitivity in %) retrospective MRI scoring
Histology
Retrospective rating MRI
Lesion type
n
3 T DIR
7 T DIR
3 T FLAIR
7 T FLAIR
3 T T2
7 T T2
3 T T1
7 T T1
3 T T2
7 T T2
Type I
Type II
Type III
Type IV
I–IV
WML
Total
6
27
28
22
83
16
99
6
8
12
19
45
15
60
6
6
14
20
46
15
61
5
2
13
18
38
14
52
5
9
16
20
50
15
65
5
2
7
17
31
9
40
5
2
22
18
47
15
62
5
4
14
20
43
13
56
5
5
13
17
40
15
55
4
3
10
16
33
14
47
6
5
15
20
46
12
58
(100)
(30)
(43)
(86)
(54)
(94)
(61)
(100)
(22)
(50)
(91)
(55)
(94)
(62)
(83)
(7)
(46)
(82)
(46)
(88)
(53)
(83)
(33)
(57)
(91)
(60)
(94)
(66)
(83)
(7)
(25)
(77)
(37)
(56)
(40)
(83)
(7)
(79)
(82)
(57)
(94)
(63)
(83)
(15)
(50)
(91)
(52)
(81)
(57)
Differences in lesion count between 3 T and 7 T that are statistically significant (P 5 0.05) are in bold font. WML = white matter lesion.
(83)
(19)
(46)
(77)
(48)
(94)
(56)
(67)
(11)
(36)
(73)
(40)
(88)
(47)
(100)
(19)
(54)
(91)
(55)
(75)
(59)
Post-mortem MS cortical lesion detection with 7 T MRI
marked as false positive on more than one sequence).
Taking this overlap into account, we identified a total of
20 unique false positive lesions on 7 T and 16 on 3 T
(Supplementary Table 1). Most of the MRI abnormalities
that were marked as false positive on 7 T MRI, reflected
incomplete demyelination or partial remyelination of the
grey matter (n = 8), or white matter (n = 4). The other
false positive lesions were based on diffuse white matter
(n = 4), blood vessels (n = 2), and in two cases no pathological substrate could be identified based on the PLP stain
alone.
Retrospective lesion detection
On retrospective scoring, when lesion location was revealed
to the MRI reader, a mean of 61 lesions per pulse sequence
was found on 7 T, an increase of 61% in comparison to
prospective scoring. At 3 T this increase was 117% (mean
of 52 lesions per pulse sequence). Nevertheless, in the retrospective scoring, 7 T MRI detected 18% more lesions than
3 T MRI, which was significant (Z = 2.02, P = 0.043). At 7
T, DIR, FLAIR, T2 and T2 detected more lesions than their
corresponding 3 T sequences. This was only significant for
T2 (62 versus 40; Z = 2.23, P = 0.026).
Discussion
This is the first study that histopathologically verifies the
increased multiple sclerosis cortical lesion detection using a
multi-contrast protocol at 7 T, in a uniquely large postmortem cohort. Our results indicate that the use of ultrahigh field 7 T MRI means a step forward in the detection
of cortical lesions in multiple sclerosis.
More cortical lesions were detected
at 7 T than at 3 T
At 7 T, we numerically detected more cortical lesions than
at 3 T, for all pulse sequences investigated. Results were
only statistically significant for FLAIR and T2 , with the
largest improvement seen with FLAIR, where 225% more
cortical lesions were detected at 7 T. This is comparable to
our earlier in vivo study, which showed an increase of
238% with 7 T FLAIR (De Graaf et al., 2013). Other
study groups also reported improved cortical lesion detection at 7 T in an in vivo setting, when compared to 1.5 T
and 3 T, mostly using T2 sequences (Kollia et al., 2009;
Metcalf et al., 2010; Tallantyre et al., 2010, 2011; Nielsen
et al., 2012; Sinnecker et al., 2012; De Graaf et al., 2013).
Since the introduction of MRI, the number of detected
white matter lesions in multiple sclerosis patients has improved by increasing magnetic field strength (Kilsdonk et
al., 2012). However, our results show that moving beyond
3 T does not seem to further improve sensitivity of white
matter lesion detection, as was suggested before (De Graaf
et al., 2013).
BRAIN 2016: 139; 1472–1481
| 1477
Sensitivity at 7 T is dependent on
cortical lesion type
Our results show that sensitivity of 7 T MRI is different
for each cortical lesion subtype. Figure 2 shows that, at
7 T we were able to distinguish between different cortical lesions (type I–IV) as first described by Bø et al.
(2003). This indicates a clear improvement compared
to previous 1.5 T studies (Kangarlu et al., 2007; Kollia
et al., 2009).
Most interest goes out to lesions extending from the pia
downwards into the cortex, i.e. type III and IV subpial
lesions. Subpial lesion can be extensive (Kutzelnigg et al.,
2005), they are related to the more progressive forms of the
disease (Kutzelnigg and Lassmann, 2006), as well as to
higher physical disability and worse cognitive performance
(Mainero et al., 2009; Nielsen et al., 2013). Furthermore,
very recently in this journal, it was suggested that the outer
cortical layers are most severely affected and that cortical
pathological processes are driven from the pial surface
(Barkhof and Geurts, 2015; Mainero et al., 2015). At 7
T, Mainero et al. (2015) used T2 maps to investigate quantitative differences throughout the cortex. They detected—
in vivo—a gradient in the expression of cortical pathology
in multiple sclerosis, across disease stages, which was
related to clinical disability.
We found a maximum sensitivity for type IV subpial lesions of 68%, attained with 7 T T2-weighted images. Type
III subpial lesions have always been problematic to detect
as they are located in the upper sparsely myelinated cortical
layers that create little MRI contrast. Nevertheless, they are
the most common type of grey matter lesion in multiple
sclerosis, and indeed dominated our current sample
(Peterson et al., 2001; Bø et al., 2003; Geurts et al.,
2005). In our study at 7 T, type III lesion detection rate
remains poor-to-moderate, but the T2 images stand out
with a sensitivity of 32%, whereas at 3 T, the maximum
sensitivity for type III subpial lesions was only 11% (T2).
An example of increased type III lesion detection with 7 T
is displayed in Fig. 3.
Previous in vivo studies also showed increased detection
of subpial lesions, with use of T2 sequences at 7 T (Kollia
et al., 2009; Mainero et al., 2009; Pitt et al., 2010; Nielsen
et al., 2012). One of those studies detected about five times
more cortical lesions with T2 than with 3 T DIR, and
hence suggested the use of T2 at 7 T to be the new ‘gold
standard’ for cortical lesion detection (Nielsen et al., 2012).
However, using a multi-contrast protocol at 7 T verified by
histopathology, we have reached a different conclusion: optimal sensitivity for cortical lesions was attained with
FLAIR and T2-weighted sequences, whereas T2 provides
a relative benefit when type III lesions are concerned.
From these results we cautiously deduce that there is no
‘optimal’ 7 T sequence for the detection of grey matter
lesions, as none of the five sequences involved detected significantly more lesions than any other sequence. Therefore,
1478
| BRAIN 2016: 139; 1472–1481
I. D. Kilsdonk et al.
Figure 2 Different types of cortical lesions as detected with 7 T T2 MRI.
Figure 3 Type III lesion detected at 7 T FLAIR and 7 T T2 , but not at 3 T. T2 = T2 -weighted. The figure shows a type IV lesion as
well, which is visible at all 7 T and 3 T images.
we propose that although increasing field strength has a
clear effect on grey matter lesion detection, the particular
type of sequence that is chosen at ultrahigh field (7 T) may
be of less importance.
Our sample included a low number of type I lesions, but
this lesion type is readily visible on MRI. Sensitivity of 7 T
for the detection of type I lesions was very high, up to even
100% with T2. Likewise, at 3 T, high detection for type I
lesions was achieved with DIR (83%), as was shown in
previous research (Geurts et al., 2005; Nielsen et al.,
2012). The type II lesions in our sample were barely visible
with MRI, only 11% at 7 T (FLAIR and T2) and 4% at 3
T (DIR, T2), which is probably caused by the small size of
this type of lesion.
There have been two other 7 T post-mortem MRI studies
in multiple sclerosis, which found slightly higher sensitivities for cortical lesions than the present study. We found a
prospective sensitivity for cortical lesions of up to 35%
with our 7 T T2-weighted sequence, whereas 42% with
white matter-attenuated inversion recovery turbo field-
Post-mortem MS cortical lesion detection with 7 T MRI
echo, and 46% and 47% with T2 were reported (Pitt et
al., 2010; Yao et al., 2014). Retrospective sensitivities of
93% (white matter-attenuated inversion recovery turbo
field-echo), 82%, and 65% (T2 ) were found (Pitt et al.,
2010; Yao et al., 2014), and in our study 60% with
FLAIR. A reason for the different sensitivities found is possibly the difference in sample sizes: only two and three
multiple sclerosis patients were studied in previous studies
(Pitt et al., 2010; Yao et al., 2014), whereas we included 19
multiple sclerosis patients. Furthermore, spatial resolutions
were slightly higher in these studies and no distinction between cortical lesion subtype was made in one study (Yao
et al., 2014).
Even at 7 T MRI, a substantial part of
cortical lesions remain undetected
After the abovementioned histopathological verification of
increased cortical lesion detection at 7 T, the question remains what proportion of the actual number of cortical
lesions we see, and—more importantly—how many we
still miss. MRI detection of purely intracortical lesions
has to date been disappointingly low. At 1.5 T, up to
95% of the intracortical lesions go undetected on MRI
(Geurts et al., 2005). When Geurts et al. (2005) compared
1.5 T with 4.7 T, sensitivity was still below 10% for both
field strengths (Geurts et al., 2008). Grey matter-specific
sequences such as 3D DIR raised prospective sensitivity
to 18%, which means that only the proverbial ‘tip of the
iceberg’ is visible, even at higher fields (Seewann et al.,
2011, 2012).
Although our prospective sensitivity rates at ultra-high
field harbour greater promise, we realize that (subpial) cortical pathology is more extensive than what ultra-high field
7 T images reveal. Even retrospectively, still 40% of all
cortical lesions were missed. Why not all cortical multiple
sclerosis lesions are visible remains unclear. Studies reporting that size of the cortical lesions is one of the major
predictors of MRI visibility (Pitt et al., 2010; Seewann et
al., 2011) are supported by the fact that in our present 7 T
study a higher resolution, and hence improved capability to
BRAIN 2016: 139; 1472–1481
| 1479
see smaller details, led to improved cortical lesion detection.
Varying degrees of inflammatory activity within lesions
might also affect lesion visibility on MRI.
In the retrospective analysis, lesion detection rates at 7 T
and 3 T converged; at 3 T there was a higher advantage of
unblinding from the histopathological results. This also
means that training and upfront optimization of 3 T sequences is required. Even retrospectively, the detection of
type II lesions remains a problem, due to the small size of
these lesions. The sensitivity for detection of subpial cortical lesions in a retrospective manner is very high, for type
III and IV lesions sensitivities of 79% (T2 ) and 91% was
achieved, respectively (DIR, FLAIR and T2-weighted).
Our observation that some cortical hyperintensities that
were detected on MRI did not appear to be lesions upon
histopathological analysis (Fig. 4), but rather incomplete
demyelination or partial remyelination should be subject
of future research. The majority of MRI sequences are sensitive to detecting tissue alterations in the brain parenchyma, but not very specific. This means that
differentiating the good (remyelination) from the bad (demyelination) remains problematic, even at 7 T. The discovery of an imaging marker for remyelination would be of
help in monitoring treatment effects. Previously, opportunities for magnetic resonance spectroscopy or PET imaging
have been described with regard to this issue (Barkhof,
1997), but ultra-high field imaging could probably also
be of help in finding a more specific imaging marker for
remyelination (Schmierer et al., 2009). Furthermore, in
future studies it would be interesting to investigate the
MRI-only white matter lesions with a more suitable staining for remyelination, such as Luxol Fast Blue-Periodic
Acid Schiff.
Limitations
In our study, pulse sequences were individually optimized
at each field strength, leading to a difference in spatial
resolution and acquisition time. The advantages of
moving to ultra-high field without decreasing image quality, because of the increased signal-to-noise ratio at 7 T,
Figure 4 Partially remyelinated type I lesion on 7 T T2 -weighted image.
1480
| BRAIN 2016: 139; 1472–1481
would otherwise not have been fully exploited.
Furthermore, a post-mortem study setting might not be
fully representative of the in vivo multiple sclerosis population, however results of cortical lesion detection are similar (De Graaf et al., 2013). In general, in the post-mortem
setting we can obtain higher resolution images than in the
in vivo setting, due to an (almost) unrestricted acquisition
time. Furthermore, movement artefacts are almost absent.
Because T1 and T2 relaxation times change due to formalin
fixation in post-mortem tissue (Pfefferbaum et al., 2004),
this may result in different contrast and cortical lesion detection compared to in vivo images. Nonetheless, the relative contrast (between white matter and grey matter, and
between lesion and white matter/grey matter) is spared,
allowing post-mortem observations to correlate qualitatively well with the in vivo observations.
As the number of cortical lesions per subtype was relatively small, statistical testing of differences between subtypes of lesions could not be performed, and results were
presented in a descriptive fashion. Furthermore, the fact
that we did not find one sequence at 7 T that was
‘better’ than the others, might be related to the power of
the study.
Conclusion
In conclusion, at ultra-high field 7 T MRI, the use of a
multi-contrast protocol more than doubles detection of cortical lesions in multiple sclerosis, compared to 3 T MRI.
Setting aside the fact that the clinical implementation of 7 T
scanners might be still a bridge too far, the increased detection of cortical lesions with 7 T MRI in multiple sclerosis patients can be of high clinical relevance, as they are
difficult to visualize at standard field and are nonetheless
important for understanding physical and cognitive disability. Including cortical lesions in the diagnostic criteria of
multiple sclerosis—at present based on white matter lesions
only—(Polman et al., 2011) might further increase accuracy
of establishing the diagnosis (Filippi et al., 2010). Still, we
realize that (subpial) cortical pathology is more extensive
than what even ultra-high field 7 T images can reveal.
Acknowledgements
The authors would like to thank the Netherlands Brain
Bank for providing the brain tissue and the VUmc
Pathology department and the MS-MRI autopsy team for
their help in acquiring the data.
Funding
This work was supported by a grant provided by the Dutch
MS Research Foundation (grant no. 09-358b and EU-FP7)
and the Noaber Foundation (Lunteren, The Netherlands).
I. D. Kilsdonk et al.
Supplementary material
Supplementary material is available at Brain online.
References
Barkhof F, Geurts JJG. Laminar cortical damage in multiple sclerosis.
Brain 2015; 138: 828–9.
Barkhof F. Imaging of remyelination. Mult Scler 1997; 3: 129–32.
Bø L, Vedeler CA, Nyland HI, Trapp BD, Mørk SJ. Subpial demyelination in the cerebral cortex of multiple sclerosis patients.
J Neuropathol Exp Neurol 2003; 62: 723–32.
Calabrese M, Gallo P. Magnetic resonance evidence of cortical onset
of multiple sclerosis. Mult Scler 2009; 15: 933–41.
Daams M, Geurts JJG, Barkhof F. Cortical imaging in multiple sclerosis: recent findings and ‘grand challenges’. Curr Opin Neurol 2013;
26: 345–52.
Filippi M, Evangelou N, Kangarlu A, Inglese M, Mainero C, Horsfield
MA, et al. Ultra-high-field MR imaging in multiple sclerosis.
J Neurol Neurosurg Psychiatry 2014; 85: 60–6.
Filippi M, Rocca MA, Calabrese M, Sormani MP, Rinaldi F, Perini P,
et al. Intracortical lesions: relevance for new MRI diagnostic criteria
for multiple sclerosis. Neurology 2010; 75: 1988–94.
Geurts JJG, Blezer ELA, Vrenken H, van der Toorn A, Castelijns
JA, Polman CH, et al. Does high-field MR imaging improve cortical lesion detection in multiple sclerosis? J Neurol 2008; 255:
183–91.
Geurts JJG, Bö L, Pouwels PJW, Castelijns JA, Polman CH, Barkhof F.
Cortical lesions in multiple sclerosis: combined postmortem MR
imaging and histopathology. Am J Neuroradiol 2005; 26: 572–7.
Geurts JJG, Pouwels PJW, Uitdehaag BMJ, Polman CH, Barkhof F,
Castelijns JA. Intracortical lesions in multiple sclerosis: improved
detection with 3D double inversion-recovery MR imaging.
Radiology 2005; 236: 254–60.
De Graaf WL, Kilsdonk ID, Lopez-Soriano A, Zwanenburg JJM,
Visser F, Polman CH, et al. Clinical application of multi-contrast
7-T MR imaging in multiple sclerosis: Increased lesion detection
compared to 3 T confined to grey matter. Eur Radiol 2013; 23:
528–40.
Kangarlu A, Bourekas EC, Ray-Chaudhury A, Rammohan KW.
Cerebral cortical lesions in multiple sclerosis detected by MR imaging at 8 Tesla. Am J Neuroradiol 2007; 28: 262–6.
Kidd D, Barkhof F, McConnell R, Algra PR, Allen IV, Revesz T.
Cortical lesions in multiple sclerosis. Brain 1999; 122: 17–26.
Kilsdonk ID, de Graaf WL, Barkhof F, Wattjes MP. Inflammation
high-field magnetic resonance imaging. Neuroimaging Clin N Am
2012; 22: 135–57.
Kilsdonk ID, de Graaf WL, Soriano AL, Zwanenburg JJ, Visser F,
Kuijer JPA, et al. Multicontrast MR imaging at 7T in multiple sclerosis: highest lesion detection in cortical gray matter with 3D-FLAIR.
Am J Neuroradiol 2013; 34: 791–6.
Kollia K, Maderwald S, Putzki N, Schlamann M, Theysohn JM, Kraff
O, et al. First clinical study on ultra-high-field MR imaging in patients with multiple sclerosis: comparison of 1.5T and 7T. Am J
Neuroradiol 2009; 30: 699–702.
Kutzelnigg A, Lassmann H. Cortical demyelination in multiple sclerosis:
a substrate for cognitive deficits? J Neurol Sci 2006; 245: 123–6.
Kutzelnigg A, Lucchinetti CF, Stadelmann C, Brück W, Rauschka H,
Bergmann M, et al. Cortical demyelination and diffuse white matter
injury in multiple sclerosis. Brain 2005; 128: 2705–12.
Lucchinetti CF, Popescu BFG, Bunyan RF, Moll NM, Roemer SF,
Lassmann H, et al. Inflammatory cortical demyelination in early
multiple sclerosis. N Engl J Med 2011; 365: 2188–97.
Mainero C, Benner T, Radding A, van der Kouwe A, Jensen R, Rosen
BR, et al. In vivo imaging of cortical pathology in multiple sclerosis
using ultra-high field MRI. Neurology 2009; 73: 941–8.
Post-mortem MS cortical lesion detection with 7 T MRI
Mainero C, Louapre C, Govindarajan ST, Giannı̀ C, Nielsen AS, CohenAdad J, et al. A gradient in cortical pathology in multiple sclerosis by
in vivo quantitative 7 T imaging. Brain 2015; 138: 932–45.
Metcalf M, Xu D, Okuda DT, Carvajal L, Srinivasan R, Kelley DAC,
et al. High-resolution phased-array MRI of the human brain at 7
tesla: initial experience in multiple sclerosis patients. J Neuroimaging
2010; 20: 141–7.
Nielsen AS, Kinkel RP, Madigan N, Tinelli E, Benner T, Mainero C.
Contribution of cortical lesion subtypes at 7T MRI to physical and
cognitive performance in MS. Neurology 2013; 81: 641–9.
Nielsen AS, Kinkel RP, Tinelli E, Benner T, Cohen-Adad J, Mainero C.
Focal cortical lesion detection in multiple sclerosis: 3 Tesla DIR versus
7 Tesla FLASH-T2 . J Magn Reson Imaging 2012; 35: 537–42.
Peterson JW, Bö L, Mörk S, Chang A, Trapp BD. Transected neurites,
apoptotic neurons, and reduced inflammation in cortical multiple
sclerosis lesions. Ann Neurol 2001; 50: 389–400.
Pfefferbaum A, Sullivan EV, Adalsteinsson E, Garrick T, Harper C.
Postmortem MR imaging of formalin-fixed human brain.
Neuroimage 2004; 21: 1585–95.
Pitt D, Boster A, Pei W, Wohleb E, Jasne A, Zachariah CR, et al.
Imaging cortical lesions in multiple sclerosis with ultra-high-field
magnetic resonance imaging. Arch Neurol 2010; 67: 812–8.
Polman CH, Reingold SC, Banwell B, Clanet M, Cohen JA, Filippi M,
et al. Diagnostic criteria for multiple sclerosis: 2010 revisions to the
‘McDonald criteria’. Ann Neurol 2011; 69: 292–302.
Rasbash J, Charlton C, Browne WJ, Healy M, Cameron B. MLwiN
Version 2.1. Centre for Multilevel Modelling, University of Bristol 2009.
Roosendaal SD, Moraal B, Pouwels PJW, Vrenken H, Castelijns JA,
Barkhof F, et al. Accumulation of cortical lesions in MS: relation
with cognitive impairment. Mult Scler 2009; 15: 708–14.
Schmierer K, Parkes HG, So P-W. Direct visualization of remyelination
in multiple sclerosis using T2-weighted high-field MRI. Neurology
2009; 72: 472
Seewann A, Roosendaal SD, Wattjes MP, Valk PVD, Barkhof F,
Polman CH. Postmortem verification of MS cortical lesion detection
with 3D DIR. Neurology 2012; 78: 302–8.
BRAIN 2016: 139; 1472–1481
| 1481
Seewann A, Vrenken H, Kooi E, Valk PVD, Knol DL, Polman CH,
et al. Imaging the tip of the iceberg: visualization of cortical lesions
in multiple sclerosis. Mult Scler 2011; 17: 1202–10.
Sethi V, Yousry TA, Muhlert N, Ron M, Golay X, Wheeler-Kingshott
C, et al. Improved detection of cortical MS lesions with phasesensitive inversion recovery MRI. J Neurol Neurosurg Psychiatry
2012; 83: 877–82.
Simon B, Schmidt S, Lukas C, Gieseke J, Träber F, Knol DL, et al.
Improved in vivo detection of cortical lesions in multiple sclerosis
using double inversion recovery MR imaging at 3 Tesla. Eur Radiol
2010; 20: 1675–83.
Sinnecker T, Mittelstaedt P, Dörr J, Pfueller CF, Harms L, Niendorf T,
et al. Multiple sclerosis lesions and irreversible brain tissue damage:
a comparative ultrahigh-field strength magnetic resonance imaging
study. Arch Neurol 2012; 69: 739–45.
Tallantyre EC, Dixon JE, Donaldson I, Owens T, Morgan PS,
Morris PG, et al. Ultra-high-field imaging distinguishes MS lesions from asymptomatic white matter lesions. Neurology 2011;
76: 534–9.
Tallantyre EC, Morgan PS, Dixon JE, Al-Radaideh A, Brookes MJ,
Morris PG, et al. 3 Tesla and 7 Tesla MRI of multiple sclerosis
cortical lesions. J Magn Reson Imaging 2010; 32: 971–7.
Vigeveno RM, Wiebenga OT, Wattjes MP, Geurts JJG, Barkhof
F. Shifting imaging targets in multiple sclerosis: from inflammation to neurodegeneration. J Magn Reson Imaging 2012; 36:
1–19.
Wattjes MP, Barkhof F. High field MRI in the diagnosis of multiple sclerosis: high field-high yield? Neuroradiology 2009; 51:
279–92.
Wattjes MP, Lutterbey GG, Gieseke J, Träber F, Klotz L, Schmidt S,
et al. Double inversion recovery brain imaging at 3T: diagnostic
value in the detection of multiple sclerosis lesions. Am J
Neuroradiol 2007; 28: 54–9.
Yao B, Hametner S, van Gelderen P, Merkle H, Chen C, Lassmann H,
et al. 7 Tesla magnetic resonance imaging to detect cortical pathology in multiple sclerosis. PLoS One 2014; 9: e108863