New insights into migraine pathophysiology
Margarita Sanchez-del-Rioa, Uwe Reuterb and Michael A. Moskowitzc
Purpose of review
This article will review new and exciting developments in
migraine research, with particular emphasis on mutations
associated with familial hemiplegic migraine and the role of
cortical spreading depression in its pathophysiology and
treatment.
Recent findings
The recent discovery of multiple point mutations in familial
hemiplegic migraine has led to the suggestion that migraine
and its variants may be due to a paroxysmal disturbance in
ion-translocating mechanisms. Mutations associated with
familial hemiplegic migraine render the brain more
susceptible to prolonged cortical spreading depression
caused by either excessive synaptic glutamate release or
decreased removal of glutamate and potassium from the
synaptic cleft, or persistent sodium influx. Suppression of
cortical spreading depression has become an interesting
target for preventive migraine treatment. Prolonged
treatment with b-blockers, valproate, topiramate,
methysergide or amitriptyline reduced the number of
potassium-evoked cortical spreading depressions and
elevated the electrical stimulation threshold for the
induction of cortical spreading depression in rats. Recent
imaging studies in patients suffering from migraine without
aura also point to the presence of silent cortical spreading
depression as an underlying mechanism. Repeated waves
of cortical spreading depression may have deleterious
effects on brain function, and perhaps cause silent
ischaemic lesions in vulnerable brain regions such as the
cerebellum in susceptible individuals.
Summary
This review emphasizes several neurobiological aspects of
migraine that reveal paroxysmal disturbances in neuronal
and vascular function, that in turn reflect disturbances in the
maintenance of ionic gradients.
Keywords
cortical spreading depression, genetics, migraine aura,
prevention
Curr Opin Neurol 19:294–298. ß 2006 Lippincott Williams & Wilkins.
a trigeminally mediated migraine headache in humans
Department of Neurology, Hospital Ruber Internacional, Madrid, Spain,
b
Department of Neurology, Charité, Universitätsmedizin Berlin, Berlin, Germany
and cStroke and Neurovascular Regulation, Massachusetts General Hospital,
Harvard Medical School, Charlestown, Massachusetts, USA
Correspondence to Dr Michael A. Moskowitz, MD, Massachusetts General Hospital,
149 13th Street, Room 6403, Charlestown, MA 02129, USA
Tel: +1 617 726 8440; fax: +1 617 726 2547;
e-mail: Moskowitz@helix.mgh.harvard.edu
Current Opinion in Neurology 2006, 19:294–298
294
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Abbreviations
CSD cortical spreading depression
FHM familial hemiplegic migraine
ß 2006 Lippincott Williams & Wilkins
1350-7540
Introduction
Migraine is a complex disorder with multiple pheno-
types. Susceptibility is conferred by genetics and by
exposure to triggering factors, with the latter varying
considerably between patients. The neurobiology of
migraine is also complex, but considerable progress has
been made recently with the identification of specific
genetic mutations. These point mutations are linked to
severe attacks, and implicate both cortical spreading
depression (CSD) and a paroxysmal disturbance in ion-
translocating mechanisms. This article will briefly review
new and exciting developments in migraine research,
with particular emphasis on mutations associated with
familial hemiplegic migraine (FHM) and the role of CSD
in its pathophysiology and treatment.
Functional implications of gene mutations
There is growing clinical and experimental evidence that
CSD is an important pathophysiological correlate of the
migraine aura. CSD is a slowly propagating wave of
neuronal depolarization which travels across the cortex
and is followed by long-lasting suppression of neuronal
activity. CSD can be induced experimentally by cortical
trauma, high extracellular concentrations of potassium or
glutamate, inhibiton of Naþ/Kþ-ATPase, and several
other stimuli [1].
In humans, functional magentic resonance imaging
(MRI) and magnetoencephalography (MEG) have
revealed a multitude of similarities between migraine
aura and experimentally induced CSD in rodents. More-
over, CSD activates the trigeminovascular system in
experimental animals, as the migraine aura precedes
[2]. Despite this overlap, the initiation of the migraine
aura remains an enigma. Several studies using visual,
auditory and somatosensory evoked potentials indicated
interictal hyperexcitability of the human cortex in migrai-
neurs, suggesting enhanced susceptibility due to a
genetic or envireonmental cause [3,4]. In support of this
hypothesis, genetic studies in migraineurs with aura
identified several gene mutations which may explain,
 New insights into migraine pathophysiology Sanchez-del-Rio et al. 295
at least in part, enhanced neuronal susceptibility for the
initiation of CSD.
The autosomal dominant inherited subtype of migraine,
so called FHM, is characterized by long and debilitating
aura. To date, polymorphisms in at least three genes
regulating ion translocation have been implicated
(Fig. 1).
FHM type 1 (FHM-1) is caused by 17 missense
mutations in the CACNA1A gene (chromosome 19p13)
that encodes the pore-forming a1A subunit of voltage-
gated neuronal P/Q-type Ca2þ channels (Cav2.1 Ca2þ)
[5]. P/Q calcium channels are coupled to neurotransmit-
ter release and are expressed on soma and dendrites
throughout the mammalian brain, specifically in the
cerebral cortex, trigeminal ganglia and brainstem
nuclei [6]. Knock-in mice carrying the human R192Q
(Arg192!Gln) pure FHM-1 mutation reveal a gain in
function at multiple levels: (1) pure gain-of-function
effect on Ca2þ channel current, (2) enhanced neurotrans-
mission at the neuromuscular junction, and (3) reduced
threshold and increased velocity of CSD [7]. Another
mutation (S218L) in the same gene alters calcium chan-
nel activation by shifting the threshold to lower voltages,
thereby causing channel opening at more hyperpolarizing
voltages. In patients carrying the S218L mutation, minor
head trauma causes severe and prolonged aura attacks,
indicating enhanced neuronal susceptibility and pro-
longed recovery [8].
FHM type 2 (FHM-2) is caused by at least 20 different
mutations in the ATP1A2 gene (chromosome 1q23) that
Figure 1 Functional implications of gene mutations
In familial hemiplegic migraine (FHM), polymorphisms
in at least three genes regulating ion translocation FHM-1
have been implicated. FHM mutations render the brain CACNA1A gene
more susceptible to prolonged cortical spreading
depression (CSD) caused by excessive synaptic
glutamate (Glu) release (type 1) or decreased removal
Cav2.1 Ca2+
of glutamate and potassium (Kþ) from the synaptic
cleft (type 2), or by facilitating repetitive high-
frequency neuronal firing rates (type 3). Other gene
candidates may disrupt glucose uptake or utilization
and the energetics of synaptic metabolism. In all Gain of function
cases, the final consequence leads to mechanisms
that support the hyperexcitability of the migraineur
brain.
Glucose
uptake?
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
encodes the catalytic a2-subunit of a Naþ/Kþ-ATPase
pump, located mainly in glial cells [9]. Sodium pumping
provides the steep sodium gradient essential for the
transport of amino acids (such as glutamate) and calcium.
Mutations in this gene decrease pump activity and result
in a lower affinity for potassium. This causes slow clear-
ance of potassium and glutamate, which, as in FHM-1,
leads to an increased CSD susceptibility.
The third subtype of FHM (FHM-3) is caused by a
Q1489K (Gln1489!Lys) mutation in the SCN1A gene
(chromosome 2q24) encoding a voltage-gated sodium
channel [10]. Gln1489 contributes to the formation of
voltage-gated sodium channels and is important for fast
sodium-channel inactivation. Gene mutations affecting
Gln1489 eliminate fast sodium-channel inactivation, as
does the mutation in the SCN1A gene, resulting in
persistent sodium influx into the neuron. The expression
of the Q1489K mutation in human embryonic kidney
HEK293-derived cells causes accelerated recovery from
fast inactivation, as determined by electrophysiological
recording [11]. It is postulated that this recovery rate
might facilitate repetitive high-frequency neuronal firing
and increased neuronal excitability. On the other hand,
repetitive firing of neurons expressing the mutant NaV11
Q1489K channels may lead to enhanced glutamate
release and extracellular potassium concentrations,
thereby initiating and perpetuating CSD. Interestingly,
the same mutation is linked to familial simple febrile
seizures [11].
It follows from the above that FHM mutations render the
brain more susceptible to prolonged CSD caused by
FHM-2 FHM-3
ATP1A2 gene SCN1A gene
α2-subunit Sodium channel
Na+/K+ ATPase pump
Loss of function Gain of function?
↑Glu/ ↑K+
Synaptic
metabolism?
⇑ CSD
 296 Headache
either excessive synaptic glutamate release (type 1) or
decreased removal of glutamate and potassium from
the synaptic cleft (type 2), or persistent sodium influx
(type 3). Not all genes and mutations involved in hemi-
plegic migraine have been identified, however. Other
gene candidates may disrupt glucose uptake or utiliz-
ation and the energetics of synaptic metabolism [12],
which may be the case in mitochondrial disorders.
Interesting enough, single-nucleotide polymorphisms
in the insulin receptor gene have been implicated in
migraine [13]. In all cases the final consequence leads to
mechanisms that support the hyperexcitability of the
migraineur brain.
Preventive migraine therapy – inhibition of
cortical spreading depression
A host of medications are used for preventive migraine
therapy. ß-Receptor blockers (metoprolol/propranolol),
anti-epileptic agents (valproic acid/topiramate), tricyclic
anti-depressants (amitryptiline) and calcium channel
blockers (flunarizin) are widely used. The vast majority
of these drugs were designed for the treatment of dis-
orders such as hypertension or epilepsy. Their ability to
reduce the frequency and intensity of migraine attacks
was discovered by serendipity when migraineurs were
treated for co-existing disorders. The mechanism by
which they inhibit the development of migraine attacks
is largely unknown. Several mechanisms have been pro-
posed, such as blockade of 5-HT2 (5-hydroxytryptamine
type 2) receptors by ß-blockers and methysergide, or
blockade of adrenoreceptors, among others [14].
Most of the preventive migraine drugs have been
assessed in various experimental animal models, with
contradictory results. For example, methysergide and
the ß-blocker propranolol were without effect on plasma
protein extravasation or neurogenic dural vasodilation in
experimental animals [15,16]. In contrast, single-dose
valproic acid effectively aborted plasma protein extra-
vasation and stimulus-induced c-Fos expression within
the trigeminal nucleus caudalis [17]. Valproic acid
enhances inhibitory GABAergic transmission (where
GABA is g-aminobutyric acid) and reduces the concen-
trations of excitatory amino acids (for example, aspartate).
Topiramate, a new anti-epilectic drug, attenuated
neurogenic vasodilation and reduced superior sagittal
sinus-evoked firing in a dose-dependent fashion within
trigeminocervical complex neurons [18]. This substance
prevents the development of migraine attacks in humans
in a dose-dependent manner [19]. Topiramate has several
actions which are relevant, including the blockade of Naþ
and Ca2þ channels, enhancement of GABA activity,
and blockade of ionotropic glutamate receptors. It was
demonstrated recently [20] that propranolol inhibited the
responses to superior sagittal sinus stimulation in cell
bodies of thalamocortical relay neurons in the rat, and this
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
effect was mediated through ß1-adrenoceptors. It was
therefore proposed that ß1-adrenoceptor antagonists
attenuate migraine, at least in part, by actions in the
thalamus.
Most of the models used above were developed to assess
mechanisms related to acute migraine pain. Therefore it
is not surprising that most of the studies with preventive
migraine drugs in these models revealed confounding
results.
Suppression of CSD has become an interesting target for
preventive migraine treatment. This is based on the
observation that migraine headaches follow the aura
stereotypically. Halothane (1.5%) and the N-methyl-D-
aspartate (NMDA) receptor antagonist MK-801 were the
first substances shown to suppress CSD effectively, but
both are impractical or toxic for treating humans. Many
other drugs have been tested in CSD models. Single
application of flunarizine, metoprolol and valproic acid
did not attenuate CSD in adult cat or adult or immature
rat brain, as demonstrated repeatedly. In contrast, Richter
et al. [21] recently reported that topical application of the
ß-blocker propranolol (250 nmol/l–1 mmol/l) dose-
dependently reduced the speed of migration or blocked
migration of CSD. The a-receptor agonists norepi-
nephrine (noradrenaline) and clonidine were also effec-
tive, but these substances do not attenuate migraine
frequency in patients. Richter et al. [21] proposed that
ß-blockers and a-receptor agonists inhibit the cyclic
AMP-controlled release of glutamate, thereby aborting
CSD. In another study [22], single-dose treatment with
topiramate (30 mg/kg intravenous) prevented cerebral
blood flow changes and single-cell cortical spiking of
CSD in the rat and in some cats. A new drug, tonabersat,
a novel benzopyran compound, successfully inhibits
CSD, neurogenic inflammation and trigeminal nerve
stimulation-induced carotid vasodilatation. It also blocks
propagation of CSD dose-dependently, but has not been
tested in migraine prevention to date [23].
As only prolonged treatment for at least 6–8 weeks
effectively reduces the occurrence of migraine attacks
with all commercially available drugs, Ayata et al. [24]
tested the efficacy of prolonged treatment with ß-block-
ers, valproate, topiramate, methysergide and amitripty-
line on CSD in rats. All drugs, but not vehicle application,
reduced the number of potassium-induced CSDs in a
time-dependent manner, and elevated the electrical
stimulation threshold for the induction of CSD. A dose–
response curve was also obtained with topiramate and
valproic acid, indicating that higher doses had stronger
effects on CSD. Interestingly, short-term treatment
for 1 week with both drugs was without effect. These
findings are in line with clinical studies which show that
increasing topiramate and valproate doses provide greater
 New insights into migraine pathophysiology Sanchez-del-Rio et al. 297
reduction of migraine frequency and also reduce cortical
hyperexitability in migraineurs.
Although the detailed mechanism by which these pre-
ventive drugs reduce CSD has not been demonstrated,
the study by Ayata et al. [24] underscores the import-
ance of further study of cellular mechanisms and
changes in gene expression associated with CSD in
order to identify new targets for preventive migraine
therapy.
Silent cortical spreading depression in
migraine without aura
During the last few decades, compelling data have sup-
ported the theory that a phenomenon resembling CSD
underlies the migraine aura. It has also been demon-
strated in an experimental setting how cortical events can
activate the trigeminal system [2,25]. Thus, the patho-
physiology of migraine with aura is being elucidated
slowly with time. This is not so clear in migraine without
aura. There is an ongoing debate on how to explain the
activation of the trigeminovascular system in migraine
without aura, since there are no explicit aura symptoms to
clinically justify the presence of CSD.
Very recently, two major sources of data point to the
possibility of a clinically silent CSD in migraine without
aura. One of the most recent pieces of evidence is derived
from the CAMERA study [26,27]. In a Dutch population-
based MRI study a high prevalence of silent lesions was
found in the posterior circulation resembling small
infarcts. Most lesions were located in the cerebellum,
being more prevalent in migraine with aura (7.5%), but
also present to a greater extent in migraine without aura
(2.2%) as compared with controls (0.7%) [26]. Interest-
ingly, there were no cardiovascular risk factors other than
age. The location of these silent infarcts does not seem
fortuitous. On the one hand, there is clinical evidence in
migraine with and without aura of cerebellar dysfunction
in up to 83% of patients [28]. In cases of FHM (mostly
type 1, but also recently type 2), cerebellar atrophy
and hypoperfusion have been documented [29,30].
Furthermore, proton magnetic resonance spectroscopy
(1H-MRS) performed on patients with FHM-1 has docu-
mented distinct metabolic abnormalities, consistent with
decreased glutamate and N-acetylaspartate and elevated
myo-inositol in the cerebellum [31]. These findings are
consistent with neuronal impairment and glial prolifer-
ation. Therefore, there are clinical and imaging data that
support a special vulnerability of the cerebellum in
different types of migraine. CSD is known to occur not
only in the cortex, but also in the cerebellum, a structure
rich in P/Q calcium channels [32]. In cerebral ischaemia,
peri-infarct CSD-like depolarization potentiates infarct
growth, whereas preconditioning with a CSD episode
protects against subsequent ischaemic insult through
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
the induction of long-lasting molecular changes in
CSD-affected tissue [33].
Based on this evidence, one might speculate that
repeated waves of CSD in migraine patients lead to silent
infarcts in vulnerable regions, such as the cerebellum,
compared with supratentorial areas. The vulnerability
could be determined by cellular vulnerabilities (such
as high glutamate) or by cytoarchitectural differences
between the cerebellum and other regions.
The second piece of evidence for a silent CSD is pro-
vided by the recent functional study from Geraud et al.
[34]. The authors used positron emission tomography
(PET) to study seven cases of spontaneous migraine
without aura within 6 h of attack onset. In all cases a
bilateral cortical hypoperfusion was observed, together
with increased blood flow in the hypothalamus and rostral
brainstem. This study reproduces for the first time the
findings of the leading study from Woods et al. [35]. The
cortical blood flow changes resemble closely those
observed in migraine aura [36]. Thus, once more, cortical
hypoperfusion points to the presence of CSD, without an
evident clinical correlate.
Conclusion
This review has emphasized several neurobiological
aspects of migraine that reflect current ideas in migraine
genetics, imaging and pharmacology. As we continue to
apply an ever-increasing array of sophisticated tools to
patients and to animal-based studies, no doubt some of
these ideas will become expanded, while others will
require modification. The notion that many migraineurs
express paroxysmal disturbances in neuronal and vas-
cular function that reflect primary or secondary disturb-
ances in the maintenance of ionic gradients is not likely
to change.
References and recommended reading
Papers of particular interest, published within the annual period of review, have
been highlighted as:
 of special interest
 of outstanding interest
Additional references related to this topic can also be found in the Current
World Literature section in this issue (pp. 329–330).
1 Bolay H, Moskowitz MA. The emerging importance of cortical spreading
depression in migraine headache. Rev Neurol (Paris) 2005; 161:655–657.
2 Bolay H, Reuter U, Dunn AK, et al. Intrinsic brain activity triggers trigeminal
meningeal afferents in a migraine model. Nat Med 2002; 8:136–142.
3 Lang E, Kaltenhauser M, Neundorfer B, Seidler S. Hyperexcitability of the
primary somatosensory cortex in migraine – a magnetoencephalographic
study. Brain 2004; 127:2459–2469.
4 Ambrosini A, de Noordhout AM, Sandor PS, Schoenen J. Electrophysiological
studies in migraine: a comprehensive review of their interest and limitations.
Cephalalgia 2003; 23 (Suppl. 1):13–31.
5 Ophoff RA, Terwindt GM, Vergouwe MN, et al. Familial hemiplegic migraine
and episodic ataxia type-2 are caused by mutations in the Ca2þ channel gene
CACNL1A4. Cell 1996; 87:543–552.
6 Pietrobon D, Striessnig J. Neurobiology of migraine. Nat Rev Neurosci 2003;
4:386–398.
 298 Headache
7 Van den Maagdenberg AM, Pietrobon D, Pizzorusso T, et al. A Cacna1a
knockin migraine mouse model with increased susceptibility to cortical
spreading depression. Neuron 2004; 41:701–710.
8 Tottene A, Fellin T, Pagnutti S, et al. Familial hemiplegic migraine mutations
increase Ca(2þ) influx through single human CaV2.1 channels and decrease
maximal CaV2.1 current density in neurons. Proc Natl Acad Sci USA 2002;
99:13284–13289.
9 De Fusco M, Marconi R, Silvestri L, et al. Haploinsufficiency of ATP1A2
encoding the Naþ/Kþ pump alpha2 subunit associated with familial hemi-
plegic migraine type 2. Nat Genet 2003; 33:192–196.
10 Dichgans M, Freilinger T, Eckstein G, et al. Mutation in the neuronal voltage-
 gated sodium channel SCN1A in familial hemiplegic migraine. Lancet 2005;
366:371–377.
Identification in three families with FHM of a novel mutation in the SCN1A gene on
chromosome 2 that encodes voltage-gated sodium channels.
11 Mantegazza M, Gambardella A, Rusconi R, et al. Identification of an Nav1.1
sodium channel (SCN1A) loss-of-function mutation associated with familial
simple febrile seizures. Proc Natl Acad Sci USA 2005; 102:18177–18182.
12 Moskowitz MA, Bolay H, Dalkara T. Deciphering migraine mechanisms: clues
from familial hemiplegic migraine genotypes. Ann Neurol 2004; 55:276–280.
13 McCarthy LC, Hosford DA, Riley HJ. Single nucleotide polymorphism alleles
in the insulin receptor gene are associated with typical migraine. Genomics
2001; 78:135–149.
14 Cohen GL. Migraine prophylactic drugs work via ion channels. Med Hypoth-
eses 2005; 65:14–22.
15 Akerman S, Williamson DJ, Hill RG, Goadsby PJ. The effect of adrenergic
compounds on neurogenic dural vasodilatation. Eur J Pharmacol 2001;
424:53–58.
16 Markowitz S, Saito K, Moskowitz MA. Neurogenically mediated plasma
extravasation in dura mater: effect of ergot alkaloids. A possible mechanism
of action in vascular headache. Cephalalgia 1988; 8:83–91.
17 Cutrer FM, Limmroth V, Ayata G, Moskowitz MA. Attenuation by valproate of
c-fos immunoreactivity in trigeminal nucleus caudalis induced by intracisternal
capsaicin. Br J Pharmacol 1995; 116:3199–3204.
18 Akerman S, Goadsby PJ. Topiramate inhibits trigeminovascular activation: an
intravital microscopy study. Br J Pharmacol 2005; 146:7–14.
19 Diener HC, Tfelt-Hansen P, Dahlof C, et al. Topiramate in migraine prophylaxis
– results from a placebo-controlled trial with propranolol as an active control.
J Neurol 2004; 251:943–950.
20 Shields KG, Goadsby PJ. Propranolol modulates trigeminovascular re-
sponses in thalamic ventroposteromedial nucleus: a role in migraine? Brain
2005; 128:86–97.
21 Richter F, Mikulik O, Ebersberger A, Schaible HG. Nadrenergic agonists and
antagonists influence migration of cortical spreading depression in rat – a
possible mechanism of migraine prophylaxis and prevention of postischemic
neuronal damage. J Cereb Blood Flow Metab 2005; 25:1225–1235.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
22 Akerman S, Goadsby PJ. Topiramate inhibits cortical spreading depres-
sion in rat and cat: impact in migraine aura. Neuroreport 2005; 16:1383–
1387.
23 Bradley DP, Smith MI, Netsiri C, et al. Diffusion-weighted MRI used to detect
in vivo modulation of cortical spreading depression: comparison of suma-
triptan and tonabersat. Exp Neurol 2001; 172:342–353.
24 Ayata C, Jin H, Judo C, et al. Suppression of cortical spreading depression in
 migraine prophylaxis. Ann Neurol 2006; [Epub ahead of print].
This is the first study which convincingly demonstrates the efficacy of several
powerful prophylactic anti-migraine drugs to suppress the induction of CSD in a
dose and time-dependent mode. The study thereby offers a mechanism by which
preventive drugs reduce the frequency of migraine attacks.
25 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 USA
2001; 98:4687–4692.
26 Kruit MC, Launer LJ, Ferrari MD, van Buchem MA. Infarcts in the posterior
circulation territory in migraine. The population-based MRI CAMERA study.
Brain 2005; 128:2068–2077.
27 Kruit MC, van Buchem MA, Hofman PA, et al. Migraine as a risk factor for
subclinical brain lesions. JAMA 2004; 291:427–434.
28 Harno H, Hirvonen T, Kaunisto MA, et al. Subclinical vestibulocerebellar
dysfunction in migraine with and without aura. Neurology 2003; 61:1748–
1752.
29 Crawford JS, Konkol RJ. Familial hemiplegic migraine with crossed cerebellar
diaschisis and unilateral meningeal enhancement. Headache 1997; 37:590–
593.
30 Spadaro M, Ursu S, Lehmann-Horn F, et al. A G301R Naþ/Kþ-ATPase
mutation causes familial hemiplegic migraine type 2 with cerebellar signs.
Neurogenetics 2004; 5:177–185.
31 Dichgans M, Herzog J, Freilinger T, et al. 1H-MRS alterations in the cere-
bellum of patients with familial hemiplegic migraine type 1. Neurology 2005;
64:608–613.
32 Ebner TJ, Chen G. Spreading acidification and depression in the cerebellar
cortex. Neuroscientist 2003; 9:37–45.
33 Thompson CS, Hakim AM. Cortical spreading depression modifies compo-
nents of the inflammatory cascade. Mol Neurobiol 2005; 32:51–57.
34 Geraud G, Denuelle M, Fabre N, et al. Positron emission tomographic studies
 of migraine. Rev Neurol (Paris) 2005; 161:666–670.
Imaging study in spontaneous migraine without aura showing simultaneous
bilateral cortical hypoperfusion and increased blood flow in subcortical structures.
35 Woods RP, Iacoboni M, Mazziotta JC. Brief report: bilateral spreading
cerebral hypoperfusion during spontaneous migraine headache. N Engl J
Med 1994; 331:1689–1692.
36 Olesen J, Friberg L, Olsen TS, et al. Timing and topography of cerebral blood
flow, aura, and headache during migraine attacks. Ann Neurol 1990; 28:791–
798.