ORIGINAL RESEARCH

ADULT BRAIN

Juxtacortical Lesions and Cortical Thinning in Multiple Sclerosis

D. Pareto, J. Sastre-Garriga, C. Auger, Y. Vives-Gilabert, J. Delgado, M. Tintoré, X. Montalban, and A. Rovira

ABSTRACT
BACKGROUND AND PURPOSE: The role of juxtacortical lesions in brain volume loss in multiple sclerosis has not been fully clarified. The
aim of this study was to explore the role of juxtacortical lesions on cortical atrophy and to investigate whether the presence of
juxtacortical lesions is related to local cortical thinning in the early stages of MS.

MATERIALS AND METHODS: A total of 131 patients with clinically isolated syndrome or with relapsing-remitting MS were scanned on a 3T
system. Patients with clinically isolated syndrome were classified into 3 groups based on the presence and topography of brain lesions: no
lesions (n ⫽ 24), only non–juxtacortical lesions (n ⫽ 33), and juxtacortical lesions and non–juxtacortical lesions (n ⫽ 34). Patients with
relapsing-remitting MS were classified into 2 groups: only non–juxtacortical lesions (n ⫽ 10) and with non–juxtacortical lesions and
juxtacortical lesions (n ⫽ 30). A juxtacortical lesion probability map was generated, and cortical thickness was measured by using
FreeSurfer.

RESULTS: Juxtacortical lesion volume in relapsing-remitting MS was double that of patients with clinically isolated syndrome. The insula
showed the highest density of juxtacortical lesions, followed by the temporal, parietal, frontal, and occipital lobes. Patients with relapsing-
remitting MS with juxtacortical lesions showed significantly thinner cortices overall and in the parietal and temporal lobes compared with
those with clinically isolated syndrome with normal brain MR imaging. The volume of subcortical structures (thalamus, pallidum, putamen,
and accumbens) was significantly decreased in relapsing-remitting MS with juxtacortical lesions compared with clinically isolated syn-
drome with normal brain MR imaging. The spatial distribution of juxtacortical lesions was not found to overlap with areas of cortical
thinning.

CONCLUSIONS: Cortical thinning and subcortical gray matter volume loss in patients with a clinically isolated syndrome or relapsing-
remitting MS was related to the presence of juxtacortical lesions, though the cortical areas with the most marked thinning did not
correspond to those with the largest number of juxtacortical lesions.

ABBREVIATIONS: CIS ⫽ clinically isolated syndrome; CISj ⫽ CIS with juxtacortical lesion; CISn ⫽ CIS with normal brain MR imaging; CISnj ⫽ CIS without
juxtacortical lesion; JL ⫽ juxtacortical lesion; JLV ⫽ juxtacortical lesion volume; LV ⫽ lesion volume; RRj ⫽ relapsing-remitting MS with JL; RRMS ⫽ relapsing-remitting
MS; RRnj ⫽ relapsing-remitting MS without JL

M ultiple sclerosis is a chronic, persistent inflammatory-de-

myelinating disease of the central nervous system, char-
acterized pathologically by focal areas of inflammation, demy-
though focal cortical demyelinated plaques are also a promi-
nent feature, even in the earliest phases of the disease.2 Unfor-
tunately, conventional MR imaging has limited sensitivity for
elination, axonal loss, and gliosis. Brain MR imaging typically detecting cortical lesions because of their small size, the poor
shows multifocal lesions, mainly in white matter regions,1 contrast resolution, and the partial volume effects of the sub-
arachnoid spaces and surrounding cortex.3,4 Thus, histopatho-
Received March 10, 2015; accepted after revision May 8. logic studies are the only way to describe, quantify, and classify
From Unitat de Ressonància Magnètica (IDI), Servei de Radiologia (D.P., C.A., A.R.), gray matter lesions according to their position in relation to
Servei de Neurologia/Neuroimmunologia, Multiple Sclerosis Centre of Catalonia the gray-white matter surface (leukocortical or juxtacortical;
(Cemcat) (J.S.-G., M.T., X.M.), Hospital Universitari Vall d’Hebron, and Port
d’Informació Científica (Y.V.-G., J.D.), Universitat Autònoma de Barcelona, Barce- intracortical and subpial).5,6 Despite the limited sensitivity of
lona, Spain. MR imaging for detecting cortical lesions in MS, results of
Please address correspondence to Àlex Rovira, MD, Unitat de Ressonància Mag-
nètica (IDI), Servei de Radiologia, Hospital Universitari Vall d’Hebron, Passeig Vall
several studies showed that cross-sectional cortical lesion vol-
d’Hebron 119-129, 08035 Barcelona, Spain; e-mail: [email protected] ume and its increase over time are associated with progression
http://dx.doi.org/10.3174/ajnr.A4485 of disability and cognitive impairment in MS.7-10
2270 Pareto Dec 2015 www.ajnr.org
Table 1: Demographic and clinical features of the study sample 0.49 ⫻ 0.49 ⫻ 3.0 mm3); and 3) sagittal
CISn CISnj CISj RRnj RRj 3D T1-weighted gradient-echo (MPRAGE)
N (women) 24 (15) 33 (20) 34 (20) 10 (4) 30 (20) sequence (TR ⫽ 2300 milliseconds, TE ⫽
Mean (SD) age, y 36 ⫾ 10 36 ⫾ 7 35 ⫾ 8 34 ⫾ 7 36 ⫾ 7 3000 milliseconds, voxel size ⫽ 1.0 ⫻
Mean (SD) EDSS 1.4 ⫾ 0.9 1.4 ⫾ 0.9 1.9 ⫾ 1.1 3.2 ⫾ 2.0 2.5 ⫾ 1.6 1.0 ⫻ 1.2 mm3). Total scanning time for
Mean (SD) DIS 3–5 mo 3–5 mo 3–5 mo 10.4 ⫾ 8.2 y 10.6 ⫾ 7.1 y
LV mL (SD) 0 0.75 ⫾ 1.17 4.10 ⫾ 6.38 3.12 ⫾ 3.71 9.10 ⫾ 7.86 these sequences was 15 minutes.
Note:—EDSS indicates expanded disability scale; DIS, disease duration.
Image Analysis
All supratentorial JLs were identified
Brain atrophy, which is also frequently detected by MR from and manually delineated on T2-FLAIR images by using Jim soft-
the earliest stages of MS, is associated with irreversible neurologic ware (http://www.xinapse.com/home.php). JL volume (JLV) was
disability, including cognitive impairment.11-14 Whole-brain atro- also calculated. To obtain JL probability maps, T2-FLAIR images
phy has emerged as a clinically relevant component of disease and the corresponding lesion masks were normalized to the Mon-
progression, and results of several studies showed that this pa- treal Neurological Institute template by using Statistical Paramet-
rameter correlates better with disability and, in particular, with cog- ric Mapping software (SPM8; Wellcome Department of Imaging
nitive impairment than with focal lesions.15 Although most brain Neuroscience, London, U.K.; http://www.fil.ion.ucl.ac.uk/spm/
atrophy measurements are based on global or regional (gray and software/spm8). Once images and masks were spatially normal-
white matter) brain volume assessment, cortical thickness has re- ized, a mean image, which represents the JL probability map, was
cently emerged as a new way to assess cortical gray matter atrophy generated for the CIS and RRMS groups. A ROI for each brain
because decreased thickness is related to fatigue, disability in general, lobe was delineated, based on the Automatic Anatomical Labeling
and cognitive impairment in particular.13,16 This measurement set of ROIs.19 The final ROI included gray and white matter. The
seems to be dependent on focal white matter lesion volume,17 but a percentage of JLs in each brain lobe relative to the total number of
potential relationship between the presence and location of demyeli- JLs in the whole brain was calculated and normalized to the ROI
nating juxtacortical lesions (JLs) and cortical atrophy has not been volume. White matter lesion volume (LV) was estimated for each
elucidated. Therefore, the aim of this study was to explore the role of patient by using the automated Lesion Segmentation Toolbox,20
JLs on cortical atrophy and to investigate whether their presence is which obtains lesion masks and associated total LVs based on the
related to local cortical thinning in the early stages of MS. T2-FLAIR and 3D T1-weighted images. Toolbox parameters had
been optimized previously for the 2D T2-FLAIR images included
MATERIALS AND METHODS in this cohort.21 Cortical thickness was measured in single
Patients
time points with MPRAGE images by using the FreeSurfer pro-
A sample of 91 consecutive patients with a clinically isolated syn-
gram suite (version 5.1; http://surfer.nmr.mgh.harvard.edu/). 10
drome (CIS) and 40 patients with relapsing-remitting MS (RRMS)
Briefly, white matter points are chosen based on their locations in
who were undergoing brain MR imaging at Vall d’Hebron University
the Talairach space as well as on their intensity and the local
Hospital for diagnostic purposes or monitoring disease evolution
neighborhood intensities. Voxels are then classified as white mat-
were included in the study. CIS is a clinical description, and many
ter or something other than white matter based on intensity and
lesions may exist on MR imaging in these patients.18 Patients with
neighbor constraints. Cutting planes are chosen to separate the
CIS were classified into 3 groups according to the presence of JLs and
hemispheres from each other. An initial surface is then generated
other typical demyelinating lesions on brain MR examination: no
for each hemisphere by tiling the outside of the white matter mass
lesions (CISn, n ⫽ 24), only non-JLs (CISnj, n ⫽ 33), and with JLs
plus non-JLs (CISj, n ⫽ 34). Patients with RRMS were divided into 2 for that hemisphere. This initial surface is then refined to follow
groups (none of the patients with RRMS had normal MR images): the intensity gradients between the white and gray matter (this is
with only non-JLs (RRnj, n ⫽ 10) and with JLs plus non-JLs (RRj, referred to as the white surface). The white surface is then nudged
n ⫽ 30). None of the patients had only JLs in the absence of other to follow the intensity gradients between the gray matter and CSF
typical MS lesions on MR imaging. The characteristics of the 5 groups (this is the pial surface). The distance between the white and the
are summarized in Table 1. The study was approved by the local pial surface gives us the thickness at each location of cortex.22,23
ethics committee. Because the study was based on MR imaging data Subcortical gray matter volume measurements were also obtained
acquired in regular clinical practice, the need for written informed as part of the established pipeline.22,23 Estimated subcortical gray
consent from the participants was waived. matter volumes were multiplied by 100 and divided by the corre-
sponding estimated total intracranial volume obtained for each
MR Image Acquisition subject. Finally, the FreeSurfer output segmentations were care-
Images were acquired on a 3T whole-body MR scanner (Tim Trio, fully reviewed, particularly checking for accuracy at the sites
Siemens; Erlangen, Germany) with a 12-channel phased-array where JLs occurred. Cortical thickness was estimated in the whole
head coil and a whole-body transmit coil. The standard MS pro- brain, lobes, and regions, and at the vertex level in each patient.
tocol included (besides other sequences): 1) fast dual-echo T2- The mean cortical thickness value for each lobe was the average of
weighted transverse sequence (TR ⫽ 3080 milliseconds, TE1 ⫽ 21 values obtained for various FreeSurfer parcellations as follows—
milliseconds, TE2 ⫽ 91 milliseconds, voxel size ⫽ 0.78 ⫻ 0.78 ⫻ frontal lobe: caudal anterior cingulate, caudal middle frontal,
3.0 mm3); 2) transverse T2-FLAIR sequence (TR ⫽ 9000 millisec- isthmus cingulate, lateral orbitofrontal, medial orbitofrontal,
onds, TE ⫽ 87 milliseconds, TI ⫽ 2500 milliseconds, voxel size ⫽ paracentral, pars opercularis, pars orbitalis, pars triangularis, pre-
AJNR Am J Neuroradiol 36:2270 –76 Dec 2015 www.ajnr.org 2271
central, rostral anterior cingulate, superior frontal, and frontal were compared across groups by 1-way ANOVA with the factor
pole; parietal lobe: inferior parietal, pericalcarine, postcentral, group, followed by pair-wise Bonferroni post hoc comparison.
posterior cingulate, precuneus, and supramarginal; temporal Finally, to assess which disease burden–related parameter
lobe: banks of superior temporal sulcus, fusiform, entorhinal, in- (presence of JL, JLV, or LV) was more relevant in the cortical
ferior temporal, middle temporal, parahippocampal, superior thickness measurements, 3 multivariate models were tested by
temporal, temporal pole, and transverse temporal; and occipital using the presence of JL, JLV, or white matter LV, as independent
lobe: cuneus, lateral occipital, and lingual; and insula. variables. The 3 models included age and sex as covariates. Signif-
icance was set at P ⬍ .05.
Statistical Analysis
Whole white matter LV and JLV and distribution were compared
RESULTS
across groups by means of 1-way ANOVA by using the factor
Whole White Matter Lesion Volume
group, followed by pair-wise Bonferroni post hoc comparison to
The mean LV in the various patient groups is reported in Table 1.
account for multiple comparisons.
The rank order was CISnj ⬍ RRnj ⬍ CISj ⬍ RRj. The between-
Differences in cortical thickness between the groups were as-
group differences in LV were significant (P ⬍ .001), and statisti-
sessed at 3 different levels: by using the global values; by using
cally significant differences were observed for the following post
values after the parcellation process; and, at the vertex level, by
hoc comparisons (P ⬍ .002): RRj ⬎ RRnj, RRj ⬎ CISj, RRj ⬎
using the cortical thickness map. First, differences between the left
CISnj, and CISj ⬎ CISnj.
and right hemispheres were tested by 1-way ANOVA by using the
factor hemisphere, followed by post hoc Bonferroni comparison.
Juxtacortical Lesion Volume and Distribution, and
Then, the cortical thickness values obtained were compared
Probability Maps
across groups by 1-way ANOVA, followed by pair-wise Bonfer-
Mean JLV was 1.28 mL (range, 0.03–14.14 mL) in patients with
roni post hoc comparison. As to the vertex level comparison, sta-
CIS, and 3.41 mL (range, 0.10 –22.70 mL) in those with RRMS.
tistical difference maps (between patients with CISnj and patients
The absolute JLV was greater at the frontal lobe, followed by the
with CISj, and RRnj and RRj) were generated based on general
parietal, temporal, and occipital lobe, and insula in both CIS and
linear model analysis by using the FreeSurfer QDEC tool.
RRMS (Table 2). The calculation of JLV relative to the total vol-
In the estimation of subcortical gray matter volumes, the
ume of each lobe in patients with CIS showed that the insula had
right-left difference was assessed by 1-way ANOVA with the fac-
the highest volume of JLs per volume of lobe, followed by the
tor hemisphere. Differences in subcortical gray matter volumes
temporal, parietal, frontal, and occipital lobes. The rank order for
RRMS was insula, followed by temporal, frontal, parietal, and
Table 2: Distribution of JLsa occipital (Table 2).
LPMvol CIS, LPM␳ CIS, LPMvol RRMS, LPM␳ RRMS,
Lobe no. (%) % no. (%) % Global and Lobar Cortical Thickness
Frontal 11.6 (39.3) 1.7 27.0 (46) 4.0 In general, the lowest cortical thickness was measured in patients
Insula 1.3 (4.5) 5.0 1.9 (3.2) 7.0
with RRj, followed by CISj, RRnj, CISnj, and CISn at both global
Parietal 7.7 (26.2) 2.0 14.2 (24.3) 3.7
Temporal 7.7 (26.12) 2.6 13.9 (23.8) 4.7 and lobar levels (Table 3). The percentage difference between
Occipital 1.1 (3.9) 0.7 1.6 (2.8) 0.9 CISn and RRj ranged from 0% (insula) to 4%– 4.5% (occipital,
Note:—LPM indicates lesion probability map; LPMvol, absolute volume in milliliters temporal, and parietal lobes). Regions with the highest attenua-
of JL per lobe (percentage of global JL volume); LPM␳, percentage of JL volume tion of JL do not correspond to the ones with the largest thickness
relative to lobe volume.
a
Total JLV (LPMvol) in milliliters (percentage relative to global amount), and percent- loss. A visual comparison of the lobar cortical thickness loss
age relative to lobe volume in patients with CIS and patients with RRMS. (mean of the right and left hemispheres) in RRj compared with

Table 3: Global and lobar mean cortical thickness (in mm) for the groups studied
P Value P Value
Lobe (Factor Group)a CISn CISnj CISj RRnj RRj (R vs L)b
Global_R .064 2.50 (0.13)c 2.47 (0.09) 2.44 (0.16) 2.47 (0.09) 2.40 (0.13) .806
Global_L .074 2.50 (0.13) 2.48 (0.19) 2.45 (0.16) 2.49 (0.09) 2.41 (0.13)
Frontal_R .389 2.55 (0.16) 2.53 (0.16) 2.51 (0.16) 2.55 (0.11) 2.49 (0.11) .006
Frontal_L .424 2.52 (0.17) 2.47 (0.11) 2.46 (0.15) 2.50 (0.13) 2.46 (0.12)
Insula_R .944 3.01 (0.16) 3.04 (0.14) 3.02 (0.25) 2.99 (0.15) 3.03 (0.13) .711
Insula_L .881 3.05 (0.19) 3.05 (0.14) 3.01 (0.24) 3.03 (0.16) 3.02 (0.15)
Parietal_R .056 2.22 (0.11)c 2.17 (0.07) 2.15 (0.17) 2.19 (0.08) 2.12 (0.13) .414
Parietal_L .025 2.22 (0.11)c 2.20 (0.09) 2.17 (0.17) 2.21 (0.09) 2.12 (0.14)
Temporal_R .119 2.93 (0.14) 2.90 (0.17) 2.87 (0.21) 2.91 (0.21) 2.80 (0.21) .002
Temporal_L .084 3.01 (0.14) 2.97 (0.16) 2.94 (0.20) 2.96 (0.16) 2.88 (0.19)
Occipital_R .014 2.29 (0.14) 2.30 (0.11)c 2.24 (0.17) 2.29 (0.12) 2.19 (0.14) .224
Occipital_L .047 2.30 (0.13) 2.31 (0.09)c 2.28 (0.13) 2.29 (0.11) 2.22 (0.15)
Note:—R indicates right hemisphere; L, left hemisphere. Numbers in parentheses are the SD.
a
P value of the 1-way ANOVA test with group as a factor.
b
P value of the hemisphere effect (R vs L hemispheres values).
c
P ⬍ .05 compared with RRj.

2272 Pareto Dec 2015 www.ajnr.org

 and left paracentral regions. Finally, cortical thickness for CISnj
was significantly thinner compared with CISn at the right frontal
pole.

Cortical Thickness and Juxtacortical Lesion Location

Only patients with JLs (CISj and RRj) were included in this sub-
analysis. The regions analyzed were those in which a JL had been
detected in at least 9 patients: banks of superior temporal sulcus,
fusiform, inferior temporal, insula, postcentral, rostral middle
frontal, superior frontal, superior temporal, and temporal pole. In
separate analyses of the CIS and RRMS groups, the mean cortical
thickness in each selected region was compared between patients
with and without JLs (2-sample t test). In the RRMS group alone,
a nonsignificant trend to decreased cortical thickness in the supe-
rior frontal region was seen in patients with JLs (P ⫽ .079). Fi-
nally, to visually check whether changes in cortical thickness
showed co-localization with JLs, a binary mask of the JL proba-
FIG 1. Percentage cortical thickness loss in RRj relative to CISn for the
different brain lobes (frontal, insula, parietal, temporal, and occipital). bility map was transferred to the inflated brain representation and
Values were color-coded and ranged from 0%–5%. Juxtacortical le- overlaid onto the corresponding contrast (CISnj ⬎ CISj with the
sion probability map (right) for patients with CIS (cold scale) and pa- CIS map, and RRnj ⬎ RRj with the RRMS map). On visual in-
tients with RRMS (warm scale), scaled to the maximum value (10%).
Results were overlaid on the mean filled structural image of the whole spection, cortical thickness changes were not found to co-localize
cohort studied. with JL probability maps (Figs 2 and 3).

Subcortical Gray Matter Volume Changes and Presence

CISnj and the JL probability map (of patients with CIS and pa-
of Juxtacortical Lesions
tients with RRMS) can be seen in Fig 1.
The estimated total intracranial volume did not differ between
On multivariate analysis of global values, the presence of JLs
groups: mean (SD) CISn ⫽ 1337.73 ⫾ 189.85 mL, CISnj ⫽
was significant (P, adjusted R2) in the right (.004, 0.166) and left
1331.71 ⫾ 192.74 mL, CISj ⫽ 1381.25 ⫾ 173.46 mL, RRnj ⫽
(.005, 0.165) hemispheres, JLV was significant in the right hemi-
1363.95 ⫾ 190.06 mL, and RRj ⫽ 1392.59 ⫾ 184.77 mL. Subcor-
sphere (.033, 0.303), and LV was significant in the left hemisphere
tical gray matter volumes in all the structures analyzed were lower
(.039, 0.646). Lobar analysis showed that insular cortical thickness
in patients with RRj (Table 4). Volume loss ranged from 8% (cau-
was not related with the presence of JLs, JLV, or LV, in either
date) to almost 20% (accumbens) when compared with CISnj.
hemisphere. The (P, adjusted R2) right frontal (.048, 0.145; .043,
The largest right-to-left asymmetry was found in the pallidum
0.639), right parietal (.005, 0.156; .020, 0.696), and right occipital
(ratio, 0.87– 0.92), followed by the putamen (0.94 – 0.98). The
(.000, 0.157; .007, 0.753) lobes showed a significant association
caudate was also significantly asymmetrical, though in the oppo-
with the presence of JLs or LV. JLV was significant in the right
site direction: right-to-left ratio ranged between 1.03 and 1.06.
parietal (.018, 0.327) and right occipital (.011, 0.315) lobes. Left
On multivariate analysis, the presence of JLs was a significant
frontal cortical thickness was not associated with any of the fac-
variable in the right thalamus (P, adjusted R2) (.000, 0.130), left
tors; left temporal lobe thickness was associated with the presence
thalamus (.001, 0.120), right caudate (.000, 0.135), left caudate
of JLs (P, adjusted R2) (.009, 0.114) and in left parietal (.002,
(.000, 0.152), right putamen (.000, 0.162), left putamen (.000,
0.166; .035, 0.649; .005, 0.368) and left occipital lobules (.004,
0.147), left pallidum (.000, 0.145), right hippocampus (.020,
0.146; .038, 0.639; .001, 0.417), thickness was associated with the 3
0.124), left hippocampus (.046, 0.127), right amygdala
factors (presence JLs, LV, JLV).
(.004, 0.081), right accumbens (.000, 0.171) and left accumbens
(.000, 0.120); JLV and LV did not reach significance in any region.
Regional Cortical Thickness
Results at a regional level were along the same lines as the ones
obtained at a global and lobar level, with the group RRj showing DISCUSSION
the thinnest cortical thickness, followed by CISj, RRnj, CISnj, and This study investigated, for the first time, the association between
CISn. Significant (post hoc) comparisons indicate that CISn mea- the presence and topography of JLs and cortical thinning, and
sured cortical thickness was greater compared with RRj in the subcortical gray matter volume measurements in patients with
following regions (by following FreeSurfer nomenclature): right CIS and patients with RRMS.
cuneus, right entorhinal, right and left fusiform, right inferior Patients with RRMS had a larger number of JLs than patients
parietal, right lateral occipital, right and left precuneus, and left with CIS. In the cohort studied, the JL volume in patients with
paracentral. Another (post hoc) comparison that was significant RRMS was twice the volume seen in patients with CIS, but JL
was CISnj over RRj in the right cuneus, right lateral occipital, left brain distribution was similar in the 2 patient populations. Most
isthmus cingulate, left paracentral, and left transverse temporal JLs were located in the frontal lobe, followed by the parietal, tem-
areas. In addition, RRj showed a significant cortical thinning poral, and occipital lobes, a distribution that is similar to the re-
compared with CISj in the left banks of superior temporal sulcus ported distribution of pure cortical lesions in RRMS.24,25 When
AJNR Am J Neuroradiol 36:2270 –76 Dec 2015 www.ajnr.org 2273
 tients with RRj, but the value in RRnj
was even lower than in CISj, though the
differences were not significant. These
results may indicate that JL accumula-
tion is not disconnected from accumu-
lation of lesions in other areas of the
brain and that they both are part of the
same disease process regardless of dis-
ease stage.
The RRj group was found to have the
greatest global and regional cortical
thinning in areas of the temporal, pari-
etal, and occipital lobe. The most
marked thickness loss (RRj compared
with CISn) was in the right entorhinal
cortex (8%), in line with the findings of
Narayana et al.26 Entorhinal thinning is
considered a predictor of cognitive de-
cline in Alzheimer disease.27 Our study
did not include a neuropsychological
evaluation; therefore, we could not ana-
FIG 2. Inflated brain, displaying areas of reduced cortical thickness (red areas) in CISj compared lyze the value of entorhinal atrophy for
with CISnj (significance level P ⬍ .01 for display purposes). Blue areas indicate regions where predicting cognitive decline in our pa-
cortical thickness in CISnj is reduced compared with CISj. The CIS JL lesion probability map was
overlaid in purple (right column, left hemisphere; left column, right hemisphere; superior row, tients with MS. When following the en-
lateral view; inferior row, medial view). The JL lesion probability map is a binary mask for display torhinal cortex, we found greater thin-
purposes. No overlap was seen between the areas, showing significant changes in thickness and ning at the right precuneus and cuneus
the JL lesion probability map.
(6%). Decreases in the remaining re-
gions of the right brain hemisphere and
all regions of the left brain hemisphere
ranged from 3% to 5%. The precuneus
and fusiform were the only regions that
showed bilateral cortical thinning.
Our results also indicated that corti-
cal thinning was more closely associated
with the presence of JLs than with total
LV or JL volume, though LV was ⬍10
mL in 90% of patients. Results of studies
reported to date, including LV as a vari-
able, are not conclusive because the
threshold value used to define high and
low LV has not been established univo-
cally.12,14 In a previous study by Charil
et al,17 the frontal, insula, and temporal
regions showed the most marked corti-
cal thinning, and cortical atrophy corre-
lated with lesion load.
Visual inspection of JL probability
maps overlaid onto the cortical change
FIG 3. Inflated brain displaying areas of reduced cortical thickness (red areas) in RRj compared maps of the 2 hemispheres showed no
with RRnj (significance level P ⬍ .01, for display purposes). Blue areas indicate regions in which overlapping in either patients with CIS
cortical thickness in RRnj is reduced compared with RRj. The RR JL lesion probability map was or those with RRMS. A more detailed
overlaid in purple (right column, left hemisphere; left column, right hemisphere; superior row, subanalysis of this aspect confirmed that
lateral view; inferior row, medial view). The JL lesion probability map was a binary mask, just for
display purposes. No overlap was seen between the areas showing significant changes in thick- differences in cortical thickness were not
ness and the JL lesion probability map. directly related to the presence of JLs in
any specific region, which indicates that
JLVs were normalized by the lobe volume, the insula appeared to cortical thinning in MS is not closely dependent on the topogra-
be the region with the highest attenuation of JLs, followed by the phy of JLs. This discrepancy may arise from the fact that only a
temporal, parietal, and frontal lobes. The LV was higher in pa- percentage of JLs and a minority of intracortical lesions are visible
2274 Pareto Dec 2015 www.ajnr.org
Table 4: Subcortical gray matter volumes percentage relative to total intracranial volume for the groups studied
Subcortical Gray P Value P Value
Matter (Factor Group)a CISng CISnjg CISjg RRnjg RRjg (R vs L)b
f e e
Thal_R .000 0.502 (0.041) 0.486 (0.060) 0.473 (0.058) 0.477 (0.092) 0.424 (0.064) .986
Thal_L .003 0.492 (0.042)e 0.485 (0.053)e 0.474 (0.062) 0.476 (0.010) 0.432 (0.066)
Pallidum_R .029 0.115 (0.013)c 0.114 (0.016)d 0.109 (0.015) 0.110 (0.017) 0.103 (0.017) .000
Pallidum_L .003 0.132 (0.018)e 0.129 (0.017)d 0.118 (0.023) 0.124 (0.027) 0.113 (0.021)
Caudate_R .069 0.277 (0.037) 0.278 (0.036) 0.257 (0.041) 0.271 (0.052) 0.256 (0.035) .015
Caudate_L .025 0.268 (0.036) 0.266 (0.034) 0.247 (0.037) 0.259 (0.043) 0.242 (0.038)
Putamen_R .009 0.428 (0.037)d 0.427 (0.074)d 0.408 (0.072) 0.410 (0.085) 0.369 (0.072) .020
Putamen_L .002 0.452 (0.042)e 0.452 (0.069)e 0.428 (0.081) 0.430 (0.082) 0.384 (0.073)
Accum_R .001 0.054 (0.011)d 0.056 (0.011)e 0.052 (0.009) 0.057 (0.012)d 0.045 (0.013) .745
Accum_L .002 0.053 (0.007) 0.058 (0.012)e 0.052 (0.009) 0.054 (0.010) 0.047 (0.010)
Hippo_R .064 0.305 (0.026) 0.315(0.041)d 0.306 (0.040) 0.306 (0.053) 0.283 (0.053) .393
Hippo_L .121 0.301 (0.026) 0.308 (0.041) 0.303 (0.037) 0.294 (0.056) 0.282 (0.044)
Amygdala_R .005 0.129 (0.017) 0.140 (0.023)e 0.128 (0.020) 0.131 (0.025) 0.119 (0.024) .440
Amygdala_L .158 0.126 (0.020) 0.134 (0.021) 0.129 (0.019) 0.128 (0.021) 0.120 (0.028)
Note:—Thal indicates thalamus; R, right; L, left; Accum, accumbens; Hippo, hippocampus.
a
P value of the 1-way ANOVA test with group as a factor.
b
P value of the hemisphere effect (R vs L hemispheres values).
c
P ⫽ .059.
d
P ⬍ .05.
e
P ⬍ .01.
f
P ⬍ .001 compared with RRj.
g
Data are expressed as mean (SD).

on conventional MR imaging (mainly because of their size), LV as a covariate in the statistical analysis. The lack of a control group
though those that are seen highly correlate with the overall num- of healthy subjects was also a limitation of the study because compar-
ber of cortical lesions detected on a histopathologic study.27 A isons with normative values would be of great interest. Further stud-
study published by Bakshi et al28 found the there was a correlation ies with a larger group and a paired control group would be needed to
between the number of cortical and juxtacortical lesions for a given confirm the main findings of our work. In addition, identification of
region and the cortical atrophy measured in that region. Discrepancy JLs was based on visual inspection by using conventional T2-FLAIR
may arise again from the fact that we evaluated the presence and sequences, which may have underestimated the number of these le-
volume (not the number) of juxtacortical lesions, which actually rep- sions. The partial volume effect could also play a role, which affected
resent a small percentage compared with cortical lesions. the visualization of the lesions in the axial images. Cortical lesions
In the present study, the volume of subcortical gray matter struc-
are abundantly present in MS and are better detected with dedicated
tures was systematically smaller in patients with RRj than in patients
pulse sequences, such as double inversion recovery, phase-sensitive
with CISn. Although the volume of subcortical gray matter structures
inversion recovery, and high-resolution 3D MPRAGE.3,29-34 Imag-
in RRnj was systematically larger than in CISj, this post hoc compar-
ing of cortical lesions at standard clinical field strength remains sub-
ison was not significant in any of the structures analyzed. A recent
optimal even when combinations of these sequences (double inver-
study that compared subcortical gray matter volumes between CIS
sion recovery, phase-sensitive inversion recovery, and MPRAGE) are
and early RRMS14 also showed that early RRMS volumes were
used because of limitations in the associated sensitivity and repro-
smaller compared with CIS and that differences were significant in
some regions (caudate, putamen, thalamus) but not in others ducibility.35,36 Thus, although these sequences have made a major
(amygdala, accumbens); this was in contrast with the absence of dif- contribution in detecting focal cortical lesions in patients with MS
ferences in cortical structures. However, these findings were not and have provided important insights about cortical abnormalities
stratified by the presence of cortical lesions or JL. In this regard, re- and their association with clinical disability and cognitive impair-
sults from our multivariate analysis indicated that the presence of JL ment in all MS subtypes, substantial research efforts are needed be-
was associated with the volume of a number of subcortical structures, fore they can be used in the diagnostic imaging work-up in clinical
whereas JLV and LV were not. practice. Finally, the local and individual effect of brain juxtacortical
Taken together, the results reported here would indicate lesions on cortical thickness measurements should also be closely
that the presence of JL could be used as a marker of diffuse gray studied.
matter damage; it could be speculated that the appearance of
JLs is the by-product of the damaging pathologic process on-
going in gray matter of people with MS and not the other way CONCLUSIONS
around (JLs and locally related cortical lesions being the main Cortical thinning and subcortical gray matter volume loss in pa-
cause of damage to gray matter). This hypothesis should be tients with CIS or RRMS was related to the presence of JLs, though
tested in future studies. the cortical areas with the most marked thinning did not corre-
The limitations of our study include the small size of the RRnj spond to those with the largest number of JLs, which may indicate
group; this is because few patients with RRMS exclusively have le- that visible focal JLs do not completely explain, but might be used
sions other than JLs. Another limitation is that the presence of both as markers of the diffuse gray matter damage that is already pres-
JL and JLV were colinear with the LV, and we were not able to add the ent in the early phases of MS.
AJNR Am J Neuroradiol 36:2270 –76 Dec 2015 www.ajnr.org 2275
ACKNOWLEDGMENTS 15. Bermel RA, Bakshi R. The measurement and clinical relevance of
We thank the “Red Española de Esclerosis Múltiple)” (RD07/ brain atrophy in multiple sclerosis. Lancet Neurol 2006;5:158 –70
CrossRef Medline
0060; RD12/0032), which is sponsored by the Fondo de Investi-
16. Ramasamy DP, Benedict RH, Cox JL, et al. Extent of cerebellum,
gación Sanitaria, the Instituto de Salud Carlos III, the Ministry of subcortical and cortical atrophy in patients with MS: a case-control
Economy and Competitiveness in Spain, and the “Ajuts per donar study. J Neurol Sci 2009;282:47–54 CrossRef Medline
Suport als Grups de Recerca de Catalunya (2009 SGR 0793),” 17. Charil A, Dagher A, Lerch JP, et al. Focal cortical atrophy in multiple
which is sponsored by the “Agència de Gestió d’Ajuts Universita- sclerosis: relation to lesion load and disability. Neuroimage 2007;34:
509 –17 CrossRef Medline
ris i de Recerca” of the Generalitat de Catalunya in Spain.
18. Miller DH, Chard DT, Ciccarelli O. Clinically isolated syndromes.
Lancet Neurol 2012;11:157– 69 CrossRef Medline
Disclosures: Jaume Sastre-Garriga—UNRELATED: Consultancy: Almirall; Grants/ 19. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated
Grants Pending: Genzyme; Payment for Lectures (including service on speakers
anatomical labeling of activations in SPM using a macroscopic an-
bureaus): Genzyme, Merck-Serono, Novartis, Teva, Biogen; Payment for Develop-
atomical parcellation of the MNI MRI single-subject brain. Neuro-
ment of Educational Presentations: Novartis; Travel/Accommodations/Meeting
Expenses Unrelated to Activities Listed: Novartis. Cristina Auger—UNRELATED: Pay- image 2002;15:273– 89 CrossRef Medline
ment for Lectures (including service on speakers bureaus): Novartis. Mar Tintoré— 20. Schmidt P, Gaser C, Arsic M, et al. An automated tool for detection
UNRELATED: Board Membership: Biogen, Genzyme, TEVA; Grants/Grants Pending: of FLAIR-hyperintense white-matter lesions in multiple sclerosis.
Biogen,* Novartis*; Payment for Lectures (including service on speakers bureaus): Neuroimage 2012;59:3774 – 83 CrossRef Medline
Biogen, Genzyme, TEVA, Novartis, Merk, Bayer. Xavier Montalban—UNRELATED: 21. Pareto D, Sastre-Garriga J, Tintore M, et al. Grey and white matter
Consultancy: Actelion, Almirall, Bayer, Biogen Idec, Genzyme, GlaxoSmithKline, segmentation in multiple sclerosis patients: correcting lesion mis-
Merck, Novartis, Sanofi-Genzyme, Roche, Teva. Alex Rovira—UNRELATED: Board classification with the lesion segmentation toolbox (LST). Mult
Membership: Genzyme, Novartis, Biogen, OLEA; Payment for Lectures (including
Scler 2013;19:(S1)410
service on speakers bureaus): Novartis, Stendhal, Biogen, Genzyme, Olea, Bracco,
22. Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I.
Bayer. *Money paid to the institution.
Segmentation and surface reconstruction. Neuroimage 1999;9:
179 –94 CrossRef Medline
REFERENCES 23. Fischl B, Dale AM. Measuring the thickness of the human cerebral
1. Filippi M, Rocca MA, Barkhof F, et al. Association between patho- cortex from magnetic resonance images. Proc Natl Acad Sci U S A
logical and MRI findings in multiple sclerosis. Lancet Neurol 2012; 2000;97:11050 –55 CrossRef Medline
11:349 – 60 CrossRef Medline 24. Battaglini M, Calabrese M, Stromillo ML, et al. Similar cortical lesion
2. Kutzelnigg A, Lucchinetti CF, Stadelmann C, et al. Cortical demyeli- distribution and cortical atrophy location in patients with relaps-
nation and diffuse white matter injury in multiple sclerosis. Brain ing-remitting multiple sclerosis. Mult Scler 2010;16:S197–352
2005;128:2705–12 CrossRef Medline 25. Calabrese M, Battaglini M, Giorgio A, et al. Imaging distribution and
3. Geurts JJ, Pouwels PJ, Uitdehaag BM, et al. Intracortical lesions in mul- frequency of cortical lesions in patients with multiple sclerosis.
tiple sclerosis: improved detection with 3D double inversion-recovery Neurology 2010;75:1234 – 40 CrossRef Medline
MR imaging. Radiology 2005;236:254 – 60 CrossRef Medline 26. Narayana PA, Govindarajan KA, Goel P, et al. Regional cortical
4. Geurts JJ, Blezer EL, Vrenken H, et al. Does high-field MR imaging thickness in relapsing remitting multiple sclerosis: a multi-center
improve cortical lesion detection in multiple sclerosis? J Neurol study Neuroimage Clin 2012;2:120 –31 CrossRef Medline
2008;255:183–91 CrossRef Medline 27. Velayudhan L, Proitsi P, Westman E, et al. Entorhinal cortex thick-
5. Wegner C, Esiri MM, Chance SA, et al. Neocortical neuronal, synaptic, ness predicts cognitive decline in Alzheimer’s disease. J Alzheimers
and glial loss in multiple sclerosis. Neurology 2006:67;960 – 67 CrossRef Dis 2013;33:755– 66 CrossRef Medline
6. Calabrese M, Favaretto A, Martini V, et al. Grey matter lesions in MS: 28. Bakshi R, Ariyaratana S, Benedict RH, et al. Fluid-attenuated inversion
from histology to clinical implications. Prion 2013;7:20 –7 CrossRef recovery magnetic resonance imaging detects cortical and juxtacorti-
Medline cal multiple sclerosis lesions. Arch Neurol 2001;58:742– 48 Medline
7. Roosendaal SD, Moraal B, Pouwels PJ, et al. Accumulation of cortical 29. Seewann A, Vrenken H, Kooi EJ, et al. Imaging the tip of the iceberg:
lesions in MS: relation with cognitive impairment. Mult Scler 2009; visualization of cortical lesions in multiple sclerosis. Mult Scler
15:708 –14 CrossRef Medline 2011;17:1202–10 CrossRef Medline
8. Calabrese M, Rocca MA, Atzori M, et al. A 3-year magnetic reso- 30. Sethi V, Yousry TA, Muhlert N, et al. Improved detection of cortical
nance imaging study of cortical lesions in relapse-onset multiple MS lesions with phase-sensitive inversion recovery MRI. J Neurol
sclerosis. Ann Neurol 2010;67:376 – 83 CrossRef Medline Neurosurg Psychiatry 2012;83:877– 82 CrossRef Medline
9. Calabrese M, Favaretto A, Poretto V, et al. Low degree of cortical 31. Calabrese M, De Stefano N, Atzori M, et al. Detection of cortical
pathology is associated with benign course of multiple sclerosis. inflammatory lesions by double inversion recovery magnetic reso-
Mult Scler 2013;19:904 –11 CrossRef Medline nance imaging in patients with multiple sclerosis. Arch Neurol 2007;
10. Filippi M, Rocca MA, Horsfield MA, et al. Imaging cortical damage 64:1416 –22 CrossRef Medline
and dysfunction in multiple sclerosis. JAMA Neurol 2013;70: 32. Wattjes MP, Lutterbey GG, Gieseke J, et al. Double inversion recov-
556 – 64 CrossRef Medline ery brain imaging at 3T: diagnostic value in the detection of multi-
11. Audoin B, Zaaraoui W, Reuter F, et al. Atrophy mainly affects the limbic ple sclerosis lesions. AJNR Am J Neuroradiol 2007;28:54 –59 Medline
system and the deep grey matter at the first stage of multiple sclerosis. 33. Simon B, Schmidt S, Lukas C, et al. Improved in vivo detection of cor-
J Neurol Neurosurg Psychiatry 2010;81:690 –95 CrossRef Medline tical lesions in multiple sclerosis using double inversion recovery MR
12. Sailer M, Fischl B, Salat D, et al. Focal thinning of the cerebral cortex imaging at 3 Tesla. Eur Radiol 2010;20:1675– 83 CrossRef Medline
in multiple sclerosis. Brain 2003;126:1734 – 44 CrossRef Medline 34. Nelson F, Poonawalla A, Hou P, et al. 3D MPRAGE improves classi-
13. Calabrese M, Rinaldi F, Mattisi I, et al. The predictive value of gray fication of cortical lesions in multiple sclerosis. Mult Scler 2008;14:
matter atrophy in clinically isolated syndromes. Neurology 2011;77: 1214 –19 CrossRef Medline
257– 63 CrossRef Medline 35. Geurts JJ, Roosendaal SD, Calabrese M, et al. Consensus recommen-
14. Bergsland N, Horakova D, Dwyer MG, et al. Subcortical and cortical dations for MS cortical lesion scoring using double inversion recov-
gray matter atrophy in a large sample of patients with clinically ery MRI. Neurology 2011;76:418 –24 CrossRef Medline
isolated syndrome and early relapsing-remitting multiple sclerosis. 36. Sethi V, Muhlert N, Ron M, et al. MS cortical lesions on DIR: not
AJNR Am J Neuroradiol 2012;33:1573–78 CrossRef Medline quite what they seem? PLoS One 2013;8:e78879 CrossRef Medline

2276 Pareto Dec 2015 www.ajnr.org