Content uploaded by Nabin Koirala
All content in this area was uploaded by Nabin Koirala on Nov 06, 2019
Content may be subject to copyright.
Interictal Hyperperfusion in the Higher Visual Cortex
in Patients With Episodic Migraine
Lars Michels, PhD; Jeanette Villanueva, Msc; Ruth O’Gorman, PhD; Muthuraman Muthuraman, PhD;
Nabin Koirala, PhD; Roman Büchler, PhD; Andreas R. Gantenbein, MD; Peter S. Sandor, MD;
Roger Luechinger, PhD; Spyros Kollias, MD; Franz Riederer, MD
Background.—Migraine pathophysiology is complex and probably involves cortical and subcortical alterations. Structural
and functional brain imaging studies indicate alterations in the higher order visual cortex in patients with migraine. Arterial
spin labeling magnetic resonance imaging (ASL-MRI) is a non-invasive imaging method for assessing changes in cerebral blood
flow (CBF) in vivo.
Objective.—To examine if interictal CBF differs between patients with episodic migraine (EM) with or without aura and
healthy controls (HC).
Methods.—We assessed interictal CBF using 2D pseudo-continuous ASL-MRI on a 3 Tesla Philips scanner (University
Hospital Zurich, Switzerland) in EM (N = 17, mean age 32.7 ± 9.9, 13 females) and HC (N = 19, mean age 31.0 ± 9.3,
Results.—Compared to HC, EM showed exclusively hyperperfusion in the right MT+ and Cohen’s d effect size was 0.99
(HC mean CBF ±SD: 33.1 ± 5.9 mL/100 g/minutes; EM mean CBF: 40.9 ± 9.4 mL/100 g/minutes). EM with aura (N = 13,
MwA) revealed hyperperfusion compared to HC in the right MT+ and superior temporal gyrus. For MT, Cohen’s d effect size
was 1.34 (HC mean CBF ± SD: 33.1 ± 5.9 mL/100 g/minutes; MwA mean CBF: 43.3 ± 8.6 mL/100 g/minutes). For the
superior temporal gyrus, Cohen’s d effect size was 1.28 (HC mean CBF ± SD: 40.1 ± 4.9 mL/100 g/minutes; MwA mean
CBF: 47.4 ± 6.4 mL/100 g/minutes). In EM, anxiety was positively associated with CBF in the parietal operculum and angular
Conclusions.—Our results suggest that extrastriate brain regions probably involved in cortical spreading depression are
associated with CBF changes in the interictal state. We conclude that ASL-MRI is a sensitive method to identify local neuro-
functional abnormalities in CBF in patients with EM in the interictal state.
Key words: migraine, episodic, cerebral blood flow, arterial spin labeling magnetic resonance imaging
Headache doi: 10.1111/head.13646
© 2019 American Headache Society Published by Wiley Periodicals, Inc.
From the Department of Neuroradiology, University Hospital Zurich, Zurich, Switzerland (L. Michels, J. Villanueva, R. Büchler,
and S. Kollias); Center for MR-Research,University Children’s Hospital Zurich, Zurich, Switzerland (L. Michels and R. O’Gorman);
University of Zurich, Zurich, Switzerland (A.R. Gantenbein, P.S. Sandor, and F. Riederer); RehaClinic Bad Zurzach, Bad Zurzach,
Switzerland (A.R. Gantenbein and P.S. Sandor); Institute for Biomedical Engineering, ETH Zurich and University of Zurich,
Zurich, Switzerland (R. Luechinger); Neurological Center Rosenhuegel and Karl Landsteiner Institute for Epilepsy Research and
Cognitive Neurology, Vienna, Austria (F. Riederer); Biomedical Statistics and Multimodal Signal Processing Unit, Movement Disorders
and Neurostimulation, Department of Neurology, Focus Program Translational Neuroscience (FTN), Johannes-Gutenberg-University
Hospital, Mainz, Germany (M. Muthuraman and N. Koirala).
Address all correspondence to L. Michels, Department of Neuroradiology, University Hospital Zurich, Sternwartstr. 6, CH-8091 Zurich,
Switzerland, email: email@example.com
Accepted for publication August 6, 2019.
Migraine is a multifactorial neurovascular disor-
der which affects about 12% of the general population1
and it is among the most disabling diseases.2,3 Migraine
is characterized by recurrent headache attacks, lasting
4-72 hours (untreated or unsuccessfully treated). The
pain is associated with at least 2 of the 4 following fea-
tures: unilateral location, pulsating quality, moderate
to strong intensity, and aggravation by routine physical
activity.4 Further, it is associated with either autonomic
signs (eg, nausea) or sensoriphobia (eg, photophobia).
In the episodic form of migraine, headache occurs on
average on less than 15days per month, whereas in the
chronic form (CM), headache occurs on ≥15/days per
month for at least three consecutive months. According
to the absence or presence of transient focal neurolog-
ical symptoms preceding or sometimes accompanying
the headache, migraine is classified as “without aura”
(MwoA) or “with aura” (MwA). MwA affects about
20-30% of migraine patients.2,5
The lack of specificity in migraine diagnosis arises,
in part, because diagnostic markers related to neuro-
biological mechanisms are lacking.6 Pain sensation
related to migraine is produced by multidirectional
and parallel processing cascades between relay stations
located in the brainstem (trigeminal nucleus caudalis)
and the brain.7,8 In analogy, the gate control theory of
pain,9,10 focused on chronic pain, postulated that neu-
ral gates in the spinal cord can be opened (or closed) by
signals descending from the brain, as well as by sensory
information ascending from the body. The centers in
the brain linked to pain processing comprise regions
associated with sensory, affective, and cognitive dimen-
sions of pain.11-14
Neuroimaging studies using perfusion weighted
imaging or positron emission tomography (PET)
identified pathophysiological changes in the dorso-
lateral prefrontal-, motor-, visual cortex, and sub-
cortical regions in patients with migraine. On the
non-neuronal level, it was shown that patients with
migraine (n = 153) demonstrated higher interictal
flow compared to healthy controls (HC) (n=2033) in
the basilar artery using 2D phase contrast imaging.15
Hemodynamic changes during migraine, indexed by
alterations in regional cerebral blood flow (CBF),
have been described using both non-invasive5,15-17
and invasive18-23 imaging techniques, such as PET.
Sanchez del Rio and colleagues examined 13 MwoA
and 6 (visual) MwA using perfusion weighted imag-
ing during spontaneous migraine episodes.17 Patients
with MwA demonstrated hypoperfusion in the visual
cortex contralateral to the hemifield defect, whereas
perfusion remained constant in MwoA. In MwoA,
an increased CBF related to pain was seen using PET
in cortical areas and in the brainstem.24 Brainstem
activation persisted after treatment of the migraine
attack with sumatriptan.14 Afridi et al25 reported
left-lateralized brainstem (dorsal pons) activation
during migraine vs the interictal state, and additional
activation in the anterior cingulate cortex, posterior
cingulate cortex, cerebellum, thalamus, insula, pre-
frontal cortex, and temporal lobes. In contrast, a
deactivation (in the migraine phase) was located in
the left pons, indicating that migraine involves brain-
stem modulations of afferent neural traffic. Using
PET, activation was observed in the periaqueductal
gray, dorsal pontine, and other midbrain regions in
episodic MwA during the premonitory phase of a
migraine.26 In MwoA, subcutaneous application of
sumatriptan did not lead to focal changes in regional
CBF during or outside of an attack, as assessed by
PET,27 indicating that triptans do not necessarily
lead to metabolic alterations in migraineurs.
In contrast to the described perfusion techniques,
advantages of arterial spin labeling magnetic resonance
imaging (ASL-MRI) include high reliability,28,29 abso-
lute quantification, and the avoidance of intravenous
contrast administration or tracers. In an ASL-MRI
case series, cerebral hyperperfusion was seen in the
frontal, parietal, or visual cortex in 3 patients during
the aura but in none of (eight) migraine patients in-
terictally.5 A single-case ASL-MRI study in a patient
without aura examined changes in CBF ictally and in-
terictally and 30minutes after oral administration of
Rizatriptan.16 CBF during the migraine showed signif-
icant relative hypoperfusion in the bilateral thalamic
and hypothalamus and hyperperfusion in the frontal
cortex compared to the migraine-free state.
There has been only one previous ASL-MRI
study that interictally measured CBF in adults with
episodic migraine (EM).30 Hodkinson and colleagues
found a significant increase in CBF in the primary
somatosensory cortex (S1), which was positively cor-
related with attack frequency. Most patients showed
signs of cutaneous allodynia, ie, skin hypersensitivity
resulting in lower pain thresholds to stimuli, which
would typically not be painful (such as hair brushing).
As S1 is part of the trigemino-cortical pathway, the
observed increase in perfusion in S1 may arise from
adaptive or maladaptive functional plasticity. The
same research group also reported hyperperfusion of
S1 in adolescents with EM31 but it was not defined if
patients showed signs of aura. ASL-MRI has also been
to characterize perfusion in acute confusional migraine
Hadjikhani and colleagues33 used high-field func-
tional MRI (fMRI) recordings during visual aura in 3
subjects.33 The authors found an initial focal increase
in the blood oxygenation level-dependent (BOLD) sig-
nal developing within extrastriate cortex (area V3A).
This BOLD change progressed across the occipital
cortex and then diminished, possibly reflecting cortical
spreading depression (CSD). The same group also re-
ported increased cortical thickness in MwA compared
to controls in areas V5 (MT+) and V3A.34 One of the
leading hypotheses in migraine pathophysiology is that
the brains of migraineurs are hyperexcitable, especially
in the extrastriate cortex.35-37 Enhanced neuronal exci-
tation results in increased extracellular potassium. If
the reuptake and other transport processes are not effi-
cient in controlling glutamate release, a wave of CSD38
is likely to arise because of this extracellular potassium
increase. CSD is a slow, self-propagating wave of neu-
ronal and glial depolarization, followed by long-lasting
suppression of neural activity. The observed functional
(Hadjikhani et al33) and structural (Granziera et al34)
abnormalities might either account for or be caused
by the hyperexcitability that triggers migraines. It has
been suggested that CSD also plays a significant role
in MwoA possibly via clinically “silent auras”39 which
is supported by the finding of cortical thickening in
V3A and V5 (MT+) in MwA and MwoA.34 It has been
demonstrated that CSD is able to activate the trigemi-
no-vascular system40 in animal experiments, explaining
the link between CSD and headache.
However, to date no study has examined if inter-
ictal changes in CBF are detectable in both episodic
migraineurs with and without aura. Using ASL-MRI,
we hypothesize that MwA will show hyperperfusion in
extrastriate visual cortex and that hyperperfusion will
be more pronounced in MwA due to a stronger mani-
festation of CSD associated with the aura.
MATERIAL AND METHODS
Design and Study Duration.—This is the primary
analysis of the reported data using a cross-sectional
design. Other imaging data (MR spectroscopy) have
been collected for all participants and results will
be presented elsewhere. No statistical power calcu-
lation was conducted prior to the study. The sample
size was based on the available data (during the study
interval) and was similar to a recent ASL study in EM
patients.30 All data were collected between December
2013 and July 2015.
Patients and Controls.—We included 17 EM and
19 HC. All EM fulfilled the ICHD-III diagnostic
criteria for EM.4 Exclusion criteria were severe psy-
chiatric disorders, cardiac problems (eg, severe
hypertension) or other neurologic disorders such as
epilepsy, stroke, traumatic brain injury, neck injury, or
cerebrovascular disease. All participants completed
prospective headache diaries, the Migraine Disabil-
ity Assessment (MIDAS)41 and Hamilton Anxiety
(HADS-A) and Depression (HADS-D) Score42 ques-
tionnaires. The HADS questionnaire comprises 7 ques-
tions for anxiety and 7 questions for depression. For
MIDAS, we assessed the attacks/month rate, ie, the
attack frequency in the last 3 months prior to the
MRI (eg, an attack frequency of 0.3 means 1 attack
in the last 3 months). Acute and prophylactic medi-
cation was recorded prior to the study interval. Apart
from migraine occurrence (days/month), we recorded
aura occurrence. Patients were free from migraine at-
tacks at least 48hours before and after the scan. The
detailed demographic data are listed in Table 1. The
study was approved by the ethics committee of can-
ton Zurich, Switzerland. All subjects provided writ-
ten informed consent prior to study enrolment. HC
received 40 Swiss Francs and patients received 50 Swiss
Francs reimbursement for their study participation. We
recruited the HC by the internal hospital webpage and
by local advertisements. We specifically asked for an
age-range of 20-50 years, in order to match to the ex-
pected age range of the migraineurs. In addition, we
used age and gender as covariates in all our CBF anal-
yses. We did not aim for a particular gender, as we had
females and males with migraine. HC were screened for
neurological disorders and other diseases according to
exclusion/exclusion criteria based on self-reports. Re-
garding the HADS questionnaire, we only included
Table 1.—Demographics and Clinical Data
HADS-A HADS-D Sex Age Sleep
EM1 9 2 f 31.3 8 0.3 7 Ye s SA
EM2 4 3 f 22.8 8 12.3 10 Ye s SA
EM3 3 2 f 29.4 9 3.3 4 Yes (vis nausea) T
EM4 2 1 f 49.2 7 1.7 16 Ye s SA,T
EM5 0 0 f 23.8 7 0.7 5 Ye s SA
EM6 10 5 f 45.0 7 2.0 5 No T
EM7 9 6 m 21.8 6 9.3 12 No T
EM8 3 5 f 49.5 8 0.3 31 Yes (vis, mot, lang) SA, opiate
EM9 7 7 f 40.5 5 8.0 27 Ye s SA
EM10 6 4 m 22.7 9 0.3 7 Ye s none
EM11 0 0 f 26.1 8 4.7 16 No SA
EM12 4 2 f 35.8 7 2.0 21 No SA
EM13 3 2 m 32.6 7 10.0 8 Ye s none
EM14 11 10 m 47.2 5 6.7 1 Ye s SA, T
EM15 13 4 f 27.7 8 1.7 16 Yes (mig sans mig) none
EM16 4 2 f 25.9 8 3.0 7 No SA
EM17 2 3 f 25.2 8 1.7 11 Ye s SA, T Riboflavin, Mg,
Mean 5.3 3.4 32.7 7.1 4.0 12.0
SD 3.9 2.6 9.9 1.2 3.8 8.3
HC1 3 3 m 30.5 9
HC2 5 1 f 20.4 1
HC3 2 0 m 52.9 7.5
HC4 9 1 f 55.0 6.5
HC5 2 1 f 30.3 7.5
HC6 6 5 m 38.2 7
HC7 3 2 m 27.1 7.5
HC8 1 2 f 29.4 6.5
HC9 2 2 m 29.7 7.5
HC10 1 0 f 24.1 6.5
HC11 2 2 m 30.0 7
HC12 3 1 f 28.6 7
HC13 2 0 m 23.3 7.5
HC14 3 0 m 39.0 6
HC15 3 1 f 33.6 7
HC16 6 2 f 25.3 6
HC17 1 1 f 24.9 7
HC18 2 1 f 25.7 8
HC19 7 1 f 25.6 6.5
Mean 3 1 31 7
SD 2.2 1.2 9.3 1.6
Between group differences
t-test 0.067 0.004 0.644 0.528
EM = episodic migraine; f = female; HC = healthy control; lang = language; m = male; mig = migraine; mot = motoric; SA = simple
analgesic; T = triptans; vis = visual.
HC with a cut-off value <11, as a score of ≥11 indi-
cates a moderate depression or anxiety,43 respectively
(see Table 1). As it is known that patients with migraine
do often shown moderate to severe signs of depres-
sion or anxiety, we did not exclude any patients with
HADS scores ≥ 11.
Anatomical Data.—Whole-brain 3D T1-weighted
structural data were recorded on a 3 Tesla MRI Philips
Ingenia scanner, equipped with a 15-element head coil.
Scanning parameters were as follows 170 slices, repe-
tition time: 8.4 ms, echo time: 3.9ms, flip angle: 8°,
voxel dimensions: 1×1×1mm, field of view: 240mm,
scan time: 4:35minutes. An experienced neuroradiolo-
gist (S.K.) examined all structural images for the pres-
ence of any brain abnormalities.
ASL Acquisition.—ASL data were acquired interictally
using a 3 Tesla Philips 2D pseudo-continuous ASL
(pCASL) sequence.44 The acquisition parameters
were: time of repetition)/time of echo = 4200/16 ms,
flip angle: 90°, FOV= 240mm, voxel size: 3×3mm,
20 slices, thickness: 6mm (no gap), imaging matrix =
80×80, labeling duration: 1.65 seconds, post-labeling
delay: 1.53seconds, SENSE factor: 2.5, scan duration
6:26 minutes. Background suppression was used with
two pulses: 1.68 and 2.76 seconds. Equilibrium brain
tissue magnetization (M0) images were recorded in a
separate run for each subject using the same parameters
as described for the pCASL sequence, apart from the
time of repetition (10,000ms).
ASL Analysis.—ASL images were preprocessed
using the toolbox ASLtbx,45 which was compatible
with MATLAB and the SPM software package (http://
www.fil.ion.ucl.ac.uk/spm/). The first step was motion
correction and denoising. Subjects were excluded from
subsequent analyses if any of the three translation
parameters exceeded half of the voxel size (ie, 3mm) or
if rotation values exceeded 1° (see Results). Denoising
included spatial smoothing with an isotropic Gaussian
filter with a full-width-at-half-maximum (FWHM) of
6mm3 to reduce inter-individual anatomical differenc-
es and further increase the signal-to-noise ratio. The
next step was pair-wise subtraction and CBF quanti-
fication using the one-compartment model.46 All CBF
images were normalized to the Montreal neurological
image (MNI) template space to allow for statistical
group comparison (see below).
CBF Quantification.—CBF was calculated on a
voxel-wise basis according to the formula:
Mcontrol−Mlabel reflects the subtraction of label and
control images, and λ = blood brain partition coeffi-
cient for water=0.9,47 T1blood=1664ms,48 τ=labeling
pulse train length = 1.68 seconds, α = labeling effi-
ciency= 0.85,44 as background suppression was used,
and w (post-tagging delay)=1.53seconds.
The labeling efficiency and the T1 of blood were
taken from literature values, derived from previous
experimental studies.44,48 The equilibrium magneti-
zation of blood was calculated from the equilibrium
magnetization of CSF, measured in 4 ROIs, multi-
plied by a correction factor for T2* decay and the
relevant blood H2O partition coefficient taken from
the literature.47 After CBF quantification, volun-
teers' mean CBF map (mL/100g/minutes) was nor-
malized to the Montreal Neurological Image (MNI)
template (average of 200 realigned brain images) to
allow for statistical between-group comparisons (see
below). The MNI template was provided by SPM12
(Wellcome Trust, UK).
Whole-Brain CBF Analysis.—For the calcula-
tion of the CBF difference images (Mcontrol–Mlabel),
simple subtraction was used because it has been
demonstrated to efficiently minimize spurious BOLD
contaminations within the CBF signal in the case of
resting-state recordings.49 Furthermore, it has been
demonstrated that simple subtraction in resting-state
CBF data works with the same performance as special
To compare the global CBF between groups,
we extracted the mean CBF for each group across
90 cortical brain regions (AAL atlas, http://neuro.
imm.dtu.dk/wiki/Autom ated_Anato mical_Labeling)
and applied unpaired two-tailed t-tests between
groups (HC vs EM and HC vs MwA). In addition,
we extracted the CBF for significant clusters show-
ing a group difference to plot and compare regional
CBF (unpaired two-tailed t-tests between groups).
For the group analysis, we used SPM12 and set up
a general linear model (GLM) in which we defined
60 ×100 ×
each group as one regressor of interest. First, we
computed F-contrasts (two-sided t-tests), to examine
if groups (“HC vs EM,” “HC (using all 19 HC) vs
MwA,” “HC vs MwoA,” and “MwA vs MwoA”) dif-
fer in CBF. Next, planned contrasts were calculated
(in case of significant F-contrasts) to examine the
directionality of CBF group differences, using inde-
pendent two-sample t-tests (one-sided with unequal
variances). For all analyses, we applied a voxel-wise
threshold of P ≤ .001 (t ≥ 3.3) with an additional
cluster-correction algorithm50 to correct for false
positives (due to the multiple comparison problem).
Recently, the validity of the applied method has
been demonstrated, ie, fMRI inferences for spatial
extent have acceptable false-positive rates.51 Based
on the ASL recording parameters, a cluster size of
k=44 was required to cluster correct the results at a
threshold of P≤.05 (cluster-corrected). Age, sex, and
global CBF were included as covariates in all analy-
ses. We subsequently ran an additional analysis with-
out including global CBF, since global CBF could
be related to regional CBF. For regions showing a
significant group difference, we computed Cohen’s d
to indicate the standardized difference between two
means (eg, HC – EM). Cohen’s d was computed as
with M2 and M1 as groups (ie, HC and EM or HC and
MwA) and SD as standard deviation. Cohen’s d effect
sizes can be small (≥0.2), medium (≥0.5), or large (≥0.8).52
We also computed separate multiple linear
regression analyses (age, sex, and global CBF were in-
cluded as covariates) between voxel-specific CBF and
clinical values, ie, migraine days, disease duration, anx-
iety, depression, and age, using a threshold of P≤.05
(cluster-corrected) to minimize the likelihood of false
positive results (due to multiple comparisons). These
variables were used as they contain migraine related
data (apart from age).
Cortical Thickness and Volume Analysis.—A corti-
cal thickness analysis was performed for all subjects
using FreeSurfer (ver.5.3.0; http://surfer.nmr.mgh
.harva rd.edu), the technical details of which are described
in prior publications.51,52 In brief, the automated
surface-based reconstruction processing stream con-
sists of skull stripping, Talairach space transformation,
gray matter (GM), white matter (WM) and cerebral spinal
fluid (CSF) boundaries optimization, segmentation of
subcortical structures, tessellation, and surface defor-
mation.53 Cortical thickness at each vertex across the
cortical mantle was calculated (in mm) as the average
distance between the GM-WM surface and GM-CSF
surface. The cortical volume (in mm3) is then comput-
ed as the product of cortical thickness and vertex area.
Anatomical labels based on Desikan-Killiany atlas
were used for parcellating cerebral cortex for obtain-
ing regional cortical thickness measurements. All
results are reported at P<.001 (uncorrected).
Demographics and Clinical data.—EM demon-
strated 4 migraine days per month (SD: 3.8; range
0.3-12.3 days) and showed moderate signs of anxi-
ety (mean HADS-A Scores 5.3 ± 3.9) and depression
(mean HADS-D-Scores 3.4±2.6). Sex, age, and average
duration of sleep did not differ between EM and HC
(Table 1). However, EM showed significantly higher
Table 2.—Whole-Brain CBF Between-Group Comparisons
Region Hemisphere MNI t Value Cluster-Size
All EM > HC MT+ (MTG) Right 60 -64 8 4.52 45
EM with aura > HC MT+ (MTG) Right 58 -64 10 6.03 159
STG Right 46-32 0 4.33 75
All results are reported at a statistical voxel threshold of P<.001 (uncorrected) with an additional cluster correction of k>44 voxels
to achieve P<.050 (cluster-corrected).
MTG=middle temporal gyrus; STG=superior temporal gyrus.
depression scores (P=.004). Only 1 patient was on pro-
phylactic medication during the study and 6 patients
took triptans as acute medication. Twelve EM had MwA
(see Table 1), although clinically it is likely that these
patients also had occasional attacks without aura.
Cerebral Blood Flow.—No data were excluded
as head motion parameters were in the range of
acceptable translation (<3mm) and rotation (<1°) and
ASL images were artefact-free. Using a Kolmogorov-
Smirnov test, we found that the global CBF was nor-
mally distributed in all HC, EM, and MwA, (P=.200,
2-tailed), warranting the use of parametric tests. As
shown in Figure 1, global cortical CBF was not sig-
nificantly different (all P > .050, independent sam-
ples 2-sided t-test) between HC (mean CBF ± SD:
32.6±3.6mL/100g/minutes), MwA (mean CBF±SD:
32.7 ± 5.1 mL/100 g/minutes), and MwoA (mean
We found a main effect of group for the F-contrasts
(“HC vs EM” and “HC vs MwA”) in the right MT+
(F=13.202 and F=13.735, respectively; P<.001). As
illustrated in Figure 2A, using a whole-brain analysis,
EM showed exclusively hyperperfusion compared to
HC in the right MT+ (unpaired 2-sample 1-sided t-test,
t=−4.52, P<.001). Cohen’s d effect size was 0.99 (HC
mean CBF ± SD: 33.1 ± 5.9 mL/100 g/minutes; EM
mean CBF: 40.9±9.4mL/100g/minutes). As shown in
Figure 2B, MwA revealed hyperperfusion in the right
MT+ and superior temporal gyrus (unpaired 2-sample
1-sided t-test, t=−6.03 and t=−4.33 with P <.001,
respectively). For MT+, Cohen’s d effect size was 1.34
(HC mean CBF±SD: 33.1 ± 5.9 mL/100g/minutes;
MwA mean CBF: 43.3±8.6 mL/100g/minutes). For
the superior temporal gyrus, Cohen’s d effect size was
1.28 (HC mean CBF±SD: 40.1±4.9mL/100g/min-
utes; MwA mean CBF: 47.4±6.4mL/100g/minutes).
A summary of the regional CBF values in the brain
areas showing a group differences is provided in Figure 3
and Table 2 for all 4 groups (HC, EM, MwA, and MwoA).
In addition, the analysis without global CBF as
nuisance variable (but including age and gender as
covariates) demonstrated – as illustrated in Figure 4 –
comparable results to the GLM including global CBF
as nuisance variable, ie, EM showed hyperperfusion in
the right MT+ (MTG) and MwA showed additional
hyperperfusion in the right STG (using the same statis-
tical thresholds) compared to HC.
MwoA (N=5) did not show hyperperfusion com-
pared to HC, even at P< .010 (uncorrected, indepen-
dent samples 2-tailed t-test). No CBF differences were
seen comparing MwA to MwoA (even at P < .01,
uncorrected, independent samples 2-tailed t-test).
In EM, anxiety was positively associated with CBF
in the left parietal operculum and right angular gyrus
(parameter estimates: 2.51+0.56 (90% confidence in-
terval) and 3.72 + 0.79, respectively). HADS-D and
MIDAS were not significantly related to CBF.
For the contrast “EM – HC,” the cortical thickness
and volume analysis did not reveal significant group
differences (all P≥.001 [uncorrected]) on the whole-
brain level. For the contrast “MwA – HC,” the volume
of the right STG was significantly larger (P < .001,
uncorrected; t-value=3.79) in MwA compared to HC.
The present study demonstrates the utility of ASL-
MRI for deriving quantitative measures of interictal
CBF in patients with migraine. We found hyperperfu-
sion in the area MT+ in migraineurs in comparison to
HC, which was even more pronounced in MwA dur-
ing the interictal state. We did not observe group dif-
ferences in the somatosensory cortex, especially S1, as
reported in a recent ASL-MRI study.30 However, most
of the patients in whom S1 perfusion changes were
observed, showed allodynia and MwA were excluded.
Fig. 1.—Illustration of global CBF in form of box-and-whisker
plots for HC, EM, and the two EM groups: MwA and MwoA.
The (bold) midline is the median of the data, with the upper
and lower limits of the box being the third and first quartile
(75th and 25th percentile), respectively. By default, the whiskers
will extend up to 1.5 times of the interquartile range. Dots that
appear outside of the whisker are outliers. Each dot represents
a single subject.
+& (0 0Z$ 0ZR$
Thus, it is unclear if CBF alterations would differ in
terms of their location and spatial extent in patients
with MwA and allodynia. In our study, most patients
showed aura symptoms. Therefore, the observed
changes in MT+ might be a feature of interictal MwA,
especially since hyperperfusion was spatially more
widespread for MwA compared to the whole sample
of patients, including MwA and MwoA.
It might appear surprising that hyperperfusion is
spatially restricted to few regions (MT+ and STG) in
Fig. 2.—Summary of whole-brain CBF between-group differences. Hyperperfusion is seen in EM compared to HC (A). Effects are
more pronounced in MwA (B). All results are reported at a statistical voxel threshold of P <.001 (uncorrected) with an additional
cluster correction of k>44 voxels to achieve a P<.050 (cluster-corrected). Age, gender, and global CBF were included nuisance
variables in the statistical GLM.
MwA. We did not find increased cortical thickness
or volume in the clusters with increased CBF, except
of a larger volume in the right STG (P<.001, uncor-
rected). Thus, CBF changes may not only be related
to anatomical changes. In contrast, a study with MwA
found cortical thickness increases which were most
pronounced in V5 and V3A.34 In addition, a recent
study revealed increased cortical thickness in V2 and
V3A in females with MwA compared to controls, and
increased cortical thickness of V2 in twins with MwA
compared to their discordant twin pairs,53 indicating
that structural changes in visual cortex may be an
inherent trait. Whether this is true for functional brain
alterations needs to be addressed in future studies. The
absence of cortical thickness and volume alterations in
visual regions in our cohort of patients may be related
to insufficient statistical power in the sample size.
Abnormal cortical excitability has been suggested to
play an important role as a possible factor predisposing
patients to the spontaneous CSD that has been suggested
to represent the pathophysiological basis of the migraine
aura.54 The pathogenetic relevance of neuronal excitabil-
ity received further support from the finding that calcium
channel structure and functions are altered in familial
hemiplegic migraine.55 The notion of increased excitabil-
ity of the visual cortex in the interictal phase in MwA
was supported by a study using a paradigm of sound-
induced flash illusions.56 In this study, illusionary effects
were decreased during the ictal and interictal phase in
MwA and only during the headache phase in MWoA.
In addition, using visual evoked potentials, it has been
demonstrated that habituation can lead to a reduction of
hyperexcitability.57,58 It is also known that MwA but not
MwoA showed significantly higher phosphene preva-
lence compared to controls, supporting the hypothesis of
a primary visual cortex hyperexcitability in MwA.59 Our
results indicate that interictal hyperexcitability is strong
in the extrastriate, especially in MT+, but not in V3A or
in the early visual cortex, seen as abnormally elevated
CBF in episodic MwoA and MwA.
In our study, interictal CBF increases in MT+ and
temporal regions were not correlated with clinical pa-
rameters. We only find a positive association between
CBF and anxiety in lateral parietal brain regions,
which are part of the default mode network. In a recent
study, Lo Buono and and colleagues (2017) reported
increased resting-state functional connectivity in MwA
compared to HC in several brain regions, including
the angular gyrus60 but no association to anxiety was
reported. The angular gyrus is part of the default mode
network, one of the main networks that are consistently
identified when an individual is at wakeful rest and not
performing an attention-demanding task. It has been
reported that ictal61 and interictal62,63 resting-state
fMRI connectivity is disturbed in the default mode
network in migraine. However, the functional meaning
of the observed correlation between anxiety and com-
ponents of the default mode network or the parietal
operculum needs to be explored in future studies.
An important question is whether the time from last
migraine attack could possibly be related to the CBF,
ie, whether CBF is dependent on the migraine cycle.
In our study, the interval to the next attack was highly
variable between participants ranging from 0.3 to 10 at-
tacks per month. In addition, we only scanned patients
Fig. 3.—Illustration of CBF in brain regions showing group
differences in form of box-and-whisker plots for each group
(HC, EM, MwA, and MwoA). (A) STG, (B) MTG. The (bold)
midline is the median of the data, with the upper and lower
limits of the box being the third and first quartile (75th and 25th
percentile), respectively. By default, the whiskers will extend up
to 1.5 times of the interquartile range. Dots that appear outside
of the whisker are outliers. Each dot represents a single subject.
+& (0 0Z$0ZR$
+& (0 0Z$0ZR$
in the interictal period, who were attack free for at least
48hours before and after MRI measurements. Therefore,
we could not perform a reasonable correlation between
CBF and time to last attack. As described, we also did
not observe a significant correlation between migraine
days (or attack frequency) and CBF.
The present study only investigated EM in the
interictal state. Possible dynamic blood flow changes
along the migraine cycle would be of great interest.
In the present study, we did not systematically screen
for allodynia, which may be associated with migraine.
Fig. 4.—Summary of whole-brain CBF between-group differences without inclusion of global CBF as nuisance variable. Hyperperfusion
is seen in EM compared to HC (A). Effects are more pronounced in MwA (B). All results are reported at a statistical voxel threshold
of P<.001 (uncorrected) with an additional cluster correction of k>44 voxels to achieve a P<.050 (cluster-corrected). Age and
gender were included a nuisance variables in the statistical GLM.
Therefore, the results may not be comparable with a
previous study in which patients frequently had al-
lodynia.30 Methodologically, we cannot address the
question whether hyperperfusion occurred because
of faster transient times or increased cerebral blood
volume in EM (and MwA). The contrast “MwA –
MwoA” did not show group differences, possibly due
to the small sample size for both groups.
We found that between-group differences were lat-
eralized to the right hemisphere. This could be related
to the side of the preferred aura and/or headache side
in the patients. Yet, we do not have information on the
dominant aura and/or headache side for all patients.
A possible influence of headache lateralization on
interictal perfusion abnormalities should be investi-
gated in further studies. In addition, we could not test
for a relation between the presence of photophobia
during attacks in patients and increased CBF com-
pared to HC in visual/extrastriatal areas, as we did
not have information on the presence of photophobia
In the interictal state, hyperperfusion was found in
the supposed region of CSD onset, located occipito-
temporally. We conclude that ASL-MRI is a sensitive
method to identify local abnormalities in CBF in epi-
sodic MwA especially, in the interictal state.
Acknowledgment: We thank Catharina Fritz-Rochner for
help with patient recruitment and data analysis. We greatly
appreciate the financial support by the Swiss Headache
Society (Hansruedi Isler Forschungss tipendium).
STATEMENT OF AUTHORSHIP:
(a) Conception and Design
Lars Michels, Andreas R. Gantenbein, Peter S.
Sandor, Spyros Kollias, Franz Riederer
(b) Acquisition of Data
Lars Michels, Jeanette Villanueva, Franz Riederer
(c) Analysis and Interpretation of Data
Lars Michels, Ruth O’Gorman, Muthuraman Muthu-
raman, Nabin Koirala, Roman Büchler, Franz
(a) Drafting the Manuscript
Lars Michels, Franz Riederer
(b) Revising It for Intellectual Content
Ruth O’Gorman, Andreas R. Gantenbein, Peter S.
Sandor, Roger Luechinger, Spyros Kollias
(a) Final Approval of the Completed Manuscript
Lars Michels, Jeanette Villanueva, Ruth O’Gorman,
Muthuraman Muthuraman, Nabin Koirala, Roman
Büchler, Andreas R. Gantenbein, Peter S. Sandor,
Roger Luechinger, Spyros Kollias, Franz Riederer
1. Manzoni GC, Stovner LJ. Epidemiology of headache.
Handb Clin Neurol. 2010;97:3-22.
2. Lipton RB, Bigal ME, Diamond M, et al. Migraine
prevalence, disease burden, and the need for preven-
tive therapy. Neurology. 2007;68:343-349.
3. Vos T, Flaxman AD, Naghavi M, et al. Years lived
with disability (YLDs) for 1160 sequelae of 289 dis-
eases and injuries 1990-2010: A systematic analysis
for the Global Burden of Disease Study 2010. Lancet.
4. Headache Classification Committee of the
International Headache Society (IHS). The Interna-
tional Classification of Headache Disorders, 3rd edi-
tion. Cephalalgia. 2018;38:1-211.
5. Pollock JM, Deibler AR, Burdette JH, et al. Migraine
associated cerebral hyperperfusion with arterial
spin-labeled MR imaging. Am J Neuroradiol. 2008;29:
6. Aguila ME, Lagopoulos J, Leaver AM, et al. Elevated
levels of GABA+ in migraine detected using (1)
H-MRS. NMR Biomed. 2015;28:890-897.
7. May A. Neuroimaging: Visualising the brain in pain.
Neurol Sci. 2007;28(Suppl. 2):S101-S107.
8. Melzack R, Wall PD. Pain mechanisms: A new theory.
9. Nathan PW. The gate-control theory of pain. A criti-
cal review. Brain. 1976;99:123-158.
10. Siegele DS. Pain and suffering. The gate control the-
ory. Am J Nurs. 1974;74:498-502.
11. Cao Y, Aurora SK, Nagesh V, Patel SC, Welch KM.
Functional MRI-BOLD of brainstem structures
during visually triggered migraine. Neurology. 2002;
12. Denuelle M, Fabre N, Payoux P, Chollet F, Geraud G.
Hypothalamic activation in spontaneous migraine
attacks. Headache. 2007;47:1418-1426.
13. Denuelle M, Fabre N, Payoux P, Chollet F, Geraud G.
Posterior cerebral hypoperfusion in migraine without
aura. Cephalalgia. 2008;28:856-862.
14. Weiller C, May A, Limmroth V, et al. Brain stem
activation in spontaneous human migraine attacks.
Nat Med. 1995;1:658-660.
15. Loehrer E, Vernooij MW, van der Lugt A, Hofman A,
Ikram MA. Migraine and cerebral blood flow in the
general population. Cephalalgia. 2015;35:190-198.
16. Kato Y, Araki N, Matsuda H, Ito Y, Suzuki C. Arterial
spin-labeled MRI study of migraine attacks treated
with rizatriptan. J Headache Pain. 2010;11:255-258.
17. Sanchez del Rio M, Bakker D, Wu O, et al. Perfusion
weighted imaging during migraine: Spontaneous vi-
sual aura and headache. Cephalalgia. 1999;19:701-707.
18. Bednarczyk EM, Wack DS, Kassab MY, et al. Brain
blood flow in the nitroglycerin (GTN) model of migraine:
Measurement using positron emission tomography and
transcranial Doppler. Cephalalgia. 2002;22:749-757.
19. Henry PY, Vernhiet J, Orgogozo JM, Caille JM.
Cerebral blood flow in migraine and cluster head-
ache. Compartmental analysis and reactivity to
anaesthetic depression. Res Clin Stud Headache.
20. Quirico PE, Allais G, Ferrando M, et al. Effects of the
acupoints PC 6 Neiguan and LR 3 Taichong on cere-
bral blood flow in normal subjects and in migraine
patients. Neurol Sci. 2014;35(Suppl. 1):129-133.
21. Olesen J. Regional cerebral blood flow and oxygen
metabolism during migraine with and without aura.
22. Andersson JL, Muhr C, Lilja A, Valind S, Lundberg
PO, Langstrom B. Regional cerebral blood flow and
oxygen metabolism during migraine with and without
aura. Cephalalgia. 1997;17:570-579.
23. Bednarczyk EM, Remler B, Weikart C, Nelson AD,
Reed RC. Global cerebral blood flow, blood volume,
and oxygen metabolism in patients with migraine
headache. Neurology. 1998;50:1736-1740.
24. Cutrer FM, O'Donnell A, Sanchez del Rio M.
Functional neuroimaging: Enhanced understanding of
migraine pathophysiology. Neurology. 2000;55:S36-S45.
25. Afridi SK, Matharu MS, Lee L, et al. A PET study
exploring the laterality of brainstem activation in
migraine using glyceryl trinitrate. Brain. 2005;128:
26. Maniyar FH, Sprenger T, Monteith T, Schankin C,
Goadsby PJ. Brain activations in the premonitory
phase of nitroglycerin-triggered migraine attacks.
27. Ferrari MD, Haan J, Blokland JA, et al. Cerebral
blood flow during migraine attacks without aura
and effect of sumatriptan. Arch Neurol. 1995;52:
28. Hodkinson DJ, Krause K, Khawaja N, et al.
Quantifying the test-retest reliability of cerebral blood
flow measurements in a clinical model of on-going
post-surgical pain: A study using pseudo-continu-
ous arterial spin labelling. Neuroimage Clin. 2013;
29. Chen Y, Wang DJ, Detre JA. Test-retest reliability of
arterial spin labeling with common labeling strategies.
J Magn Reson Imaging. 2011;33:940-949.
30. Hodkinson DJ, Veggeberg R, Wilcox SL, et al.
Primary somatosensory cortices contain altered pat-
terns of regional cerebral blood flow in the interictal
phase of migraine. PLoS ONE. 2015;10:e0137971.
31. Youssef AM, Ludwick A, Wilcox SL, et al. In child
and adult migraineurs the somatosensory cortex
stands out … again: An arterial spin labeling investi-
gation. Hum Brain Mapp. 2017;38:4078-4087.
32. Kossorotoff M, Calmon R, Grevent D, et al.
Arterial spin labeling (ASL) magnetic resonance
imaging in acute confusional migraine of childhood.
J Neuroradiol. 2013;40:142-144.
33. Hadjikhani N, Sanchez Del Rio M, Wu O, et al.
Mechanisms of migraine aura revealed by functional
MRI in human visual cortex. Proc Natl Acad Sci U S A.
34. Granziera C, DaSilva AF, Snyder J, Tuch DS,
Hadjikhani N. Anatomical alterations of the visual
motion processing network in migraine with and with-
out aura. PLoS Med. 2006;3:e402.
35. Aurora SK, Welch KM, Al-Sayed F. The threshold
for phosphenes is lower in migraine. Cephalalgia.
36. Battelli L, Black KR, Wray SH. Transcranial magnetic
stimulation of visual area V5 in migraine. Neurology.
37. Welch KM. Contemporary concepts of migraine
pathogenesis. Neurology. 2003;61:S2-S8.
38. Leao AAP. Spreading depression of activity in the
cerebral cortex. J Neurophysiol. 1944;7:359-390.
39. Pietrobon D, Striessnig J. Neurobiology of migraine.
Nat Rev Neurosci. 2003;4:386-398.
40. Lombard A, Bogdanov VB, Chauvel V, Multon S,
Schoenen J. Effects of cortical spreading depression
(CSD) on C-FOS expression in rat periaqueductal
grey matter. J Headache Pain. 2010;11:S32.
41. Stewart WF, Lipton RB, Dowson AJ, Sawyer J.
Development and testing of the Migraine Disability
Assessment (MIDAS) Questionnaire to assess head-
ache-related disability. Neurology. 2001;56:S20-
42. Zigmond AS, Snaith RP. The hospital anxiety and
depression scale. Acta Psychiatr Scand. 1983;67:361-
43. Stern AF. The hospital anxiety and depression scale.
Occup Med. 2014;64:393-394.
44. Dai W, Garcia D, de Bazelaire C, Alsop DC.
Continuous flow-driven inversion for arterial spin
labeling using pulsed radio frequency and gradient
fields. Magn Reson Med. 2008;60:1488-1497.
45. Wang Z, Aguirre GK, Rao H, et al. Empirical op-
timization of ASL data analysis using an ASL data
processing toolbox: ASLtbx. Magn Reson Imaging.
46. Buxton RB, Frank LR, Wong EC, Siewert B, Warach
S, Edelman RR. A general kinetic model for quan-
titative perfusion imaging with arterial spin labeling.
Magn Reson Med. 1998;40:383-396.
47. Herscovitch P, Raichle ME. What is the correct value
for the brain–blood partition coefficient for water?
J Cereb Blood Flow Metab. 1985;5:65-69.
48. Lu H, Clingman C, Golay X, van Zijl PC. Determining
the longitudinal relaxation time (T1) of blood at 3.0
Tesla. Magn Reson Med. 2004;52:679-682.
49. Liu TT, Wong EC. A signal processing model for
arterial spin labeling functional MRI. Neuroimage.
50. Slotnick SD, Moo LR, Segal JB, Hart J, Jr. Distinct
prefrontal cortex activity associated with item mem-
ory and source memory for visual shapes. Cogn Brain
51. Slotnick SD. Cluster success: fMRI inferences for spa-
tial extent have acceptable false-positive rates. Cogn
52. Cohen J. Statistical Power Analysis for the Behavioral
Sciences. 2nd edn. New York: Lawrence Erlbaum
53. Gaist D, Hougaard A, Garde E, et al. Migraine with
visual aura associated with thicker visual cortex.
54. Welch KM, Barkley GL, Tepley N, Ramadan NM.
Central neurogenic mechanisms of migraine. Neu-
55. Ophoff RA, Terwindt GM, Vergouwe MN, et al.
Familial hemiplegic migraine and episodic ataxia
type-2 are caused by mutations in the Ca2+ channel
gene CACNL1A4. Cell. 1996;87:543-552.
56. Brighina F, Bolognini N, Cosentino G, et al. Visual cor-
tex hyperexcitability in migraine in response to sound-
induced flash illusions. Neurology. 2015;84:2057-2061.
57. Brighina F, Palermo A, Fierro B. Cortical inhibi-
tion and habituation to evoked potentials: Relevance
for pathophysiology of migraine. J Headache Pain.
58. Coppola G, Di Lorenzo C, Schoenen J, Pierelli F.
Habituation and sensitization in primary headaches.
J Headache Pain. 2013;14:65.
59. Brigo F, Storti M, Tezzon F, Manganotti P, Nardone
R. Primary visual cortex excitability in migraine: A
systematic review with meta-analysis. Neurol Sci.
60. Lo Buono V, Bonanno L, Corallo F, et al. Functional
connectivity and cognitive impairment in migraine
with and without aura. J Headache Pain. 2017;18:72.
61. Edes AE, Kozak LR, Magyar M, et al. Spontaneous
migraine attack causes alterations in default mode
network connectivity: A resting-state fMRI case re-
port. BMC Res Notes. 2017;10:165.
62. Coppola G, Di Renzo A, Tinelli E, et al. The rest-
ing state connectivity between default-mode network
and insula encodes intensity of migraine headache.
63. Coppola G, Di Renzo A, Tinelli E, et al. Resting
state connectivity between default mode network and
insula encodes acute migraine headache. Cephalalgia.