Headache ISSN 0017-8748

C
2007 the Author doi: 10.1111/j.1526-4610.2007.00678.x
Journal compilation
C
2007 American Headache Society Published by Blackwell Publishing
Pathophysiology of HeadachePast and Present
Michael A. Moskowitz, MD, M.Sc. (Hon)
(Headache 2007;47 [Suppl 1]:S58-S63)
In 1979 my laboratory published its first article
on migraine in The Lancet, and it was entitled Neu-
rotransmitters and the fifth cranial nerve: Is there a
relation to the headache phase of migraine?
1
In this
article, we hypothesized that the trigeminal nerve was
a critical component in the pathogenesis of migraine.
At that time, the sensory innervation to the blood ves-
sels of the circle of Willis was unknown. We believed
that such a pathway, if it existed, may contain vasoac-
tive neuropeptides suchas substance Pwhichwouldbe
released into the meninges. Because hemicranial pain
was so prevalent in many migraineurs, we wondered
whether or not peptide neurotransmitters might be im-
plicatedinanipsilaterally projecting pathway fromthe
trigeminal ganglia to the meninges and its blood ves-
sels.
The headache phase of migraine may develop as a re-
sult of an abnormal interaction and perhaps abnormal
release of vasoactive neurotransmitters from terminals
of the trigeminal nerve with surrounding large intracra-
nial andextracranial bloodvessels. These bloodvessels,
which dilate during the headache phase of migraine,
are thought to receive axonal projections from all three
divisions of the trigeminal nerve. Substance P, a po-
tent vasodilating peptide, seems to be released from
From the Harvard-MIT Division of Health Science and Tech-
nology, Cambridge, MA, USA; and Stroke and Neurovascu-
lar Regulation Laboratory, Massachusetts General Hospital,
Charlestown, MA, USA.
Address all correspondence to Dr. Michael A. Moskowitz, Mas-
sachusetts General Hospital, 14913thStreet, Room6403, Stroke
and Neurovascular Regulation Laboratory, Charlestown, MA
02129-2020, USA.
trigeminal nerve endings in response to nervous stim-
ulation and is involved in the transmission of painful
stimuli within the periphery. The abnormal release of
Substance P, or as yet unidentified peptides or other
transmitters from the fifth cranial nerve, may explain
both the hemicranial pain and the vasodilation, which
are characteristics of the headache phase of migraine.
Evidence that peptides and other transmitters partici-
pate in the pathophysiology of migrainous headaches
might suggest new strategies for the prophylaxis and
treatment.
1
For the next 15 years, our labs research work fo-
cused on defining the anatomy, identifying the pep-
tide transmitters, defining the receptor populations
expressed on primary afferents, and building mod-
els, which may impact the treatment of humans. The
term trigeminovascular system was coined in 1983.
2
In 1984 we published the Neurobiology of vascular
head pain
3
in which we considered the impact of the
release of vasodilating peptides and the permeabil-
ity promoting peptide. Stimulation of the trigeminal
ganglia causes the release of peptides within the TVS.
Among trigeminal axons projecting into the meninges,
calcitonin gene related peptide (CGRP) immunoreac-
tive neurons predominate (40% of all cells), whereas
a smaller number of neurons contains substance P
(SP, 18%), nitric oxide synthase (NOS, 15%), and
other.
4
CGRPandSPpotently relax the humanmiddle
meningeal artery in vitro.
5
Even though CGRP (and
subsequent vasodilatation) does not reportedly excite
or directly sensitize meningeal nociceptors,
6
it can in-
dicate migraine attacks.
7
Possibly, the involvement of
CGRP in migraine depends upon its role as a central
neurotransmitter rather thanthroughvasodilationand
S58
Headache S59
activation of meningeal nociceptors.
8
In fact, a recent
clinical trial with a neuropeptide receptor antagonist
(CGRP) successfully aborted migraine headaches;
9
thus, CGRP is an important therapeutic target for
acute migraine treatment. We also speculated on the
possible existence of prejunctional 5-HTreceptor pop-
ulations and their importance to migraine, which also
foreshadowed the development of a 5-HT
1F
receptor
agonist.
In 1984, possible triggers of the trigeminal system
and initiators of pain were discussedas summarized
below:
Spreading depression and vascular headachesit is
reasonable to postulate that the headaches in classic mi-
graine are initiatedby activity withinthe central nervous
system. Stress, trauma, or brain injury provoke, initiate,
or relate temporally to their onset. Sleep tends to alle-
viate or terminate vascular headaches but is associated
withthe onset of cluster headaches. The brainappears to
trigger migraine headaches because cortical spreading
depression is the most common initiating event in mi-
graine attacks. Somewhat paradoxically, Leao reported
rapid and marked dilation of pial arteries and veins
shortly after the onset of electrical inactivity, which is
a part of this wave of spreading depression. Extracel-
lular concentrations of potassium reach as high as 60
mM and have been measured in tandem with the wave
of spreading depression. These increases are sufficient
to depolarize trigeminal nerve fibers surrounding pial
arteries, and could provide one mechanism linking the
aura and painful phase of migraine headaches.
3
In fact, defining events that initiate an attack by
triggering the trigeminal nerve continue as a major
focus of research today. We know that a propagating
electrophysiological event such as cortical spreading
depression (CSD), expressed clinically as an aura, de-
velops in brain.
10,11
Several studies suggest that CSDis
noxious and can trigger headaches (Fig. 1). Unfortu-
nately, auras have not been defined or captured by
high-resolution electrophysiological techniques. The
strongest evidence was provided by indirect methods
that assess flow or tissue oxygenation, albeit with high
temporal and spatial resolution. Nevertheless, the de-
polarization and release of constituents from neurons
andglia seemtobe the candidate event occurring prior
to onset of the headache and a trigger for headache.
Fig 1.Schematic showing the potential pathophysiological re-
lationships in migraine with aura. Susceptibility factors (genes)
and specific triggering events within the environment alter the
susceptibility to CSD and promote its initiation and propa-
gation. Cortical spreading depression is accompanied by re-
leased constituents from brain and blood vessels that have
been hypothesized to activate the trigeminal innervation of the
meninges. This event is facilitatedby protease alterationof basal
lamina and other membrane proteins. Within brainstem, activa-
tion of trigeminal nucleus caudalis triggers the subsequent ac-
tivation of preganglionic parasympathetic neurons and subse-
quently parasympathetic fibers innervating the meninges. The
rostral transmission of information from the nucleus caudalis
may lead to the generation of pain, whereas both the trigeminal
and parasympathetic reflex cause meningeal dilation.
Other triggers such as from the circulation or blood
vessel wall may be important as well, and not re-
quire triggering factors released from brain to cause
headache.
A few years ago, Hadjikhani and colleagues re-
ported the results of an MRI study in a patient who
could induce his own migraine attack.
12
Using high-
resolution BOLD-MRI, BOLD, the basis for func-
tional imaging, detects the ratio of non-paramagnetic
oxygenated Hb to paramagnetic deoxy-Hb. As a mea-
sure of oxygen delivery minus consumption, it is influ-
enced by neuronal activity including flow, volume and
oxygen consumption. Hadjikhani mapped changes
within occipital cortex during visual aura. She found
a marked perturbation or change in the BOLD sig-
nal, which developed sequentially along consecutive
regions of calcarine cortex. This perturbation in the
S60 April 2007
BOLD signal was identical from voxel-to-voxel along
calcarine cortex, differing only with respect to the time
of onset, beginning posteriorly and spreading anteri-
orly. We inferred from this study that migraine aura
is accompanied by a propagating brain event, which
is retinotopically related to the visual percept (mov-
ing from central to peripheral visual fields). She in-
terpreted the changes in BOLD signal as first, an in-
crease in blood flowlasting a fewminutes, followed by
a longer lasting decrease in blood flow and vasodila-
tion that dropped belowbaseline levels. This sequence
of events is similar to what is observed during cortical
spreading depression and documented repeatedly in
rodent or cat cortex.
Based on the propagation velocity (which is usu-
ally 2-3 mm per minute in the human and rat), and
the pattern of vasodilation followed by vasoconstric-
tion (characteristic of cortical spreading depression),
it seemed reasonable to conclude that an event closely
resembling cortical spreading depression was caus-
ing this patients migraine visual aura. Additionally,
spreading depression was reported to occur within the
cerebellum in experimental studies at a characteristic
rate of 2-3 mmper minute,
13
and has been identified in
hippocampus, basal ganglia, and many other silent
areas within cerebral cortex.
What is the relevance of the spreading depression
model to synaptic events that occur in the migraine
brain? First, excitatory amino acid glutamate and the
NMDA receptor are important. When injected or ap-
plied topically, excitatory amino acids trigger cortical
spreading depression. As proposed by van Harrev-
eld,
14
CSD propagation is based on glutamate release,
and in fact, glutamate itself has been shown to in-
duce SD when applied to the cortex. Since Marrannes
et al confirmed that the NMDA-receptor is an im-
portant component in the generation and depropa-
gation of SD and associated inward currents, a vari-
ety of NMDA-receptor antagonists have been shown
to block cortical SD.
1517
Moreover, CSD is trig-
gered by energy failure as, for example, in stroke
or trauma. Cortical spreading depression is triggered
by ouabain, a digitalis-like compound that inhibits
sodium-potassium ATPase, an important ion pump
ridding the cell of Na+ in exchange for K
+
. Cortical
spreading depressionalsocauses hemiparesis inexper-
imental animals that lasts for hours suggesting that it
can cause neurological symptoms.
A reason for discussing cortical spreading depres-
sion and glutamate relates to the question of how mi-
graine may develop as a consequence of genetic mu-
tations causing rare forms of migraine, and possibly
by extrapolation to more common types. Familial
hemiplegic migraine (type 1) is caused by a gain of
function mutation in the
1A
subunit of calcium chan-
nel
19
that controls glutamate release into the synaptic
cleft (so-called P/Q calcium channel). It also is asso-
ciated with mutations in subunit proteins comprising
sodium-potassium ATPase (type 2).
20
The life cycle
of a glutamate molecule at the synapse is shown in
Figure 2. Upon initiation of neuronal depolarization,
voltage-sensitive calcium channels open. This allows
calcium to enter the nerve ending, and glutamate
is then released. Glutamate binds to the postsynap-
tic membrane receptor, and then is cleared from the
synapse by the astrocyte transporters to terminate the
depolarization response.
Theastrocytehas 2important tasks. Oneis toridit-
self of glutamate, whichit does by recycling it andmov-
ing it back into the neuron. It must rid itself of sodium
by exporting it in exchange for potassium ions. If the
sodium-potassium pump fails, potassium accumulates
extracellularly, and it levels are high enough, can ini-
tiate cortical spreading depression. High intracellular
sodium prevents glutamate uptake by the transporter
because the transporter is driven by a steep sodium
gradient from outside to inside the cell. As a conse-
quence of Na, K
+
pump failure, glutamate builds up
in the synaptic cleft and extracellular potassium in-
creases. Both are associated with the initiation of CSD.
In fact, a knock-in mouse expressing the
1A
mutation
shows an increased susceptibility to CSD.
22
CSD has been further implicated in migraine
based on recently published data by Ayata and col-
leagues
23
showing that migraine prophylactic drugs
block the susceptibility to CSD triggered by 2 dif-
ferent stimuli (chemical or electrical) in 2 different
species (rats andmice). Acute treatment was not effec-
tive. Longer treatment duration corresponded to bet-
ter CSD suppression. The 5 tested drugs (topiramate,
propranolol, valproate, methysergide, and amitripty-
line) are pharmacologically and chemically distinct.
Headache S61
Fig 2.The relationship between glutamate, calcium channel, astrocytes, and the sodium potassium pump suggest how gain-of-
function and loss-of-function mutations in specific genes may increase susceptibility toward CSD(fromMoskowitz et al. Ann Neurol.
2004
18
with permission). Cav2.1 channels are mainly located in presynaptic glutamatergic terminals and regulate neurotransmitter
release as well as dendritic excitability. Presynaptic Ca-influx increases glutamate release and mutated calcium channels in FHM-1
showenhanced opening of Cav2.1 and increased glutamate release. Increased excitability in FHM-2 arises fromsodiumpump failure
particularly in the astrocytic Na-K ATPase leading to augmented extracellular levels of potassium and glutamate. Both increased
potassium and glutamate initiate cortical spreading depression.
Although the precise mechanism(s) remain for study,
these results may enable future research to identify
specific cellular and molecular targets important in the
design of more effective and safe prophylactic drugs.
Historically, there have been 2 hypotheses regard-
ing the pathogenesis of migraine. One is that migraine
originates in the brainstemand the other is that it orig-
inates in the cortex. The 2 are not mutually exclusive.
The brainstem hypothesis is supported by the concept
of a key role for descending modulation, and a change
in modulation from rostral structures, activating the
brainstemleadingtoinitiationof amigraineattack. Al-
ternatively (or perhaps in addition), migraine may be
initiated, at least in some, via a change in susceptibility
to CSD. It may turn out that CSD can be modified by
rostrally projecting pathways arising from the brain-
stem. The challenge is to establish this experimentally.
There is strengthening evidence to suggest that
CSD is a noxious event. Twelve years ago we reported
that followingCSD, theearlyimmediateresponsegene
c-Fos was activated in lamina I, II of the trigeminal
nucleus caudalis.
24
With the development of new opti-
cal imaging technology, we recently established a link
between cortical spreading depression and activation
of the trigeminal system. During CSD, blood flow and
vessel caliber significantly increased within the middle
meningeal artery, andthis increasewas duetoatrigem-
inal autonomic brain stem-dependent reflex. This re-
flex is initiated by intense neuroglial depolarization
that appears sufficient to drive trigeminal afferents
terminating within the meninges and trigeminal nu-
cleus caudalis.
25
To block this afferent autonomic re-
flex arc, we sectioned the trigeminal innervation and
the efferent parasympathetic projections to the cor-
tex. This autonomic loop may be one of the expla-
nations for parasympathetic activation during cluster
headache.
26
The precise cause for pain is not known. We
postulated that a wave of CSD moves slowly along
cortex releasing potassium, arachidonic acid, hydro-
gen ions, and nitric oxide. This leads to depolariza-
tion of the trigeminal afferent loop of the reflex arc
(Fig. 3). This in turn leads to brainstem activation.
Parasympathetic efferents and trigeminal activation
cause vasodilation, andplasma leakage withinthe dura
mater.
21
S62 April 2007
Fig 3.This illustration shows the relationship between the trigeminal nerve, meninges, brainstem, and parasympathetic efferents.
Cortical spreadingdepressionreleases constituents intothebrain, whichapproximates meningeal afferents andwhencritical levels are
reached, discharges trigeminal fibers. This causes orthodromic transmission within trigeminovascular system to activate trigeminal
nucleus caudalis and subsequently the parasympathetic efferents projecting from the sphenopalatine ganglion (Reprinted from
Iadecola et al, 2002
21
).
Several mechanisms may enable levels of released
brain and blood components to approximate and acti-
vate the trigeminovascular systemduring the aura. We
explored the possibility that CSD increased proteases
that degrade membrane proteins. During CSD, matrix
metalloproteinases (MMP) degrade laminin, collagen
type IV, a critical component of brain blood vessels.
MMP-9 is activated within 15 minutes in blood vessels
and lasts for many hours (12 hours).
27
In summary, the data suggest that migraine is a
disturbance of the most important visceral organ, the
brain. Cortical spreading depression appears to cause
migraine visual aura. There may be cases of migraine
without aura that are the consequence of CSD-like
events within brain (see Woods et al
28
). Therefore, it
is important toimprove uponthe diagnosis of migraine
and classify patients using criteria in addition to those
elicited at the bedside.
Conflict of Interest: None
REFERENCES
1. Moskowitz MA, Reinhard JFJr, Romero J, Melamed
E, Pettibone DJ. Neurotransmitters and the fifth cra-
nial nerve: Is there a relation to the headache phase
of migraine? The Lancet. 1979;2:883-885.
Headache S63
2. Liu-Chen LY, Mayberg MR, Moskowitz MA. Im-
munohistochemical evidence for a substance P-
containing trigeminovascular pathway to pial arteries
in cats. Brain Res. 1983;268:162-166.
3. Moskowitz MA. The neurobiology of vascular head
pain. Ann Neurol. 1984;16:157-168.
4. Edvinsson L. Sensory nerves in man and their role in
primary headaches. Cephalalgia. 2001 Sep;21(7):761-
4.
5. Jansen I, Uddman R, Ekman R, Olesen J, Ottos-
son A, Edvinsson L. Distribution and effects of neu-
ropeptide Y, vasoactive intestinal peptide, substance
P, and calcitonin gene-related peptide in human mid-
dle meningeal arteries: comparison with cerebral and
temporal arteries. Peptides. 1992 May-Jun;13(3):527-
36.
6. Levy D, Burstein R, Strassman AM. Calcitonin gene-
related peptide does not excite or sensitize meningeal
nociceptors: implications for the pathophysiology of
migraine. Ann Neurol. 2005 Nov;58(5):698-705.
7. Lassen LH, Haderslev PA, Jacobsen VB, Iversen HK,
Sperling B, Olesen J. CGRP may play a causative role
in migraine. Cephalalgia. 2002 Feb;22(1):54-61.
8. Strassman AM, Levy D. Response properties of dural
nociceptors in relation to headache. J Neurophysiol.
2006 Mar;95(3):1298-306.
9. Olesen J, Diener HC, Husstedt IW, et al. Calcitonin
gene-related peptide receptor antagonist BIBN 4096
BS for the acute treatment of migraine. NEngl J Med.
2004;350:1104-1110.
10. OlesenJ, LarsenB, LauritzenM. Focal hyperemiafol-
lowed by spreading oligemia and impaired activation
of rCBF in classic migraine. Ann Neurol. 1981;9:344-
352.
11. CaoY, WelchKMA, Aurora S, et al. Functional MRI-
BOLDof visually triggered headache in patients with
migraine. Arch Neurol. 1999;56:548-554.
12. Hadjikhani N, Sanchez Del RioM, WuO, et al. Mech-
anisms of migraine aura revealed by functional MRI
in human visual cortex. Proc Natl Acad Sci USA.
2001;98:4687-4692.
13. Chen G, Dunbar RL, Gao W, Ebner TJ. Role of cal-
cium, glutamate neurotransmission, and nitric oxide
in spreading acidification and depression in the cere-
bellar cortex. J Neurosci. 2001;21:9877-9887.
14. Van Harreveld A, Fifkova E. Glutamate release from
the retina during spreading depression. J Neurobiol.
1970;2:13-29.
15. Marrannes R, Willems R, De Prins E, Wauquier
A. Evidence for a role of the N-methyl-D-aspartate
(NMDA) receptor in cortical spreading depression in
the rat. Brain Res. 1988 Aug 9;457(2):226-40.
16. Lauritzen M, Hansen AJ. The effect of glutamate re-
ceptor blockade on anoxic depolarization and corti-
cal spreading depression. J Cereb Blood Flow Metab.
1992 Mar;12(2):223-9.
17. Anderson TR, Andrew RD. Spreading depression:
imaging and blockade in the rat neocortical brain
slice. J Neurophysiol. 2002 Nov;88(5):2713-25.
18. Moskowitz MA, Bolay H, Dalkara T. Deciphering
migraine mechanisms: Clues fromfamilial hemiplegic
migraine genotypes. Ann Neurol. 2004;55:276-280.
19. Ophoff RA, Terwindt GM, Vergouwe MN, et al. Fa-
milial hemiplegic migraine and episodic ataxia type 2
are caused by mutations in the calcium channel gene
CACNLIA4. Cell. 1996;87:543-552.
20. De Fusco M, Marconi R, Silvestri L, et al. Haplioin-
sufficiency of ATP1A2 encoding the Na+/K+ pump
alpha2 subunit gene is responsible for familial hemi-
plegic migraine type 2. Nat Genet. 2003;33:192-196.
21. Iadecola C. FromCSDto headache: Along and wind-
ing road. Nat Med. 2002;8:110-112.
22. vandenMaagdenbeurg AM, PietrobonD, Pizzorusso
T, et al. ACacna1a knockin migrainene mouse model
with increased susceptibility to cortical spreading de-
pression. Neuron. 2004;41:701-710.
23. Ayata C, Jin H, Kudo C, Dalkara T, Moskowitz MA.
Suppression of cortical spreading depression in mi-
graine prophylaxis. Ann Neurol. 2006 Apr;59(4):652-
61.
24. Moskowitz MA, Nozaki K, Kraig RP. Neocortical
spreading depression provokes the expression of c-
fos protein-like immunoreactivity within trigeminal
nucleus caudalis via trigeminovascular mechanisms.
J Neurosci. 1993;13:1167-1177.
25. Bolay H, Reuter U, Dunn AK, Huang Z, Boas
DA, Moskowitz MA. Intrinsic brain activity triggers
trigeminal meningeal afferents in a migraine model.
Nat Med. 2002;8:136-142.
26. Moskowitz MA. Cluster headache: Evidence for a
pathophysiologic focus in the superior pericarotid
cavernous sinus plexus. Headache. 1988;28:584-
586.
27. Gursoy-Ozdemir Y, Qiu J, Matsuoka N, et al. Cor-
tical spreading depression activates and upregulates
MMP-9. J Clin Invest. 2004;113:1447-1455.
28. Woods RP, Iacoboni M, Mazziotta JC. Brief re-
port: Bilateral spreading cerebral hypoperfusion dur-
ing spontaneous migraine headache. N Engl J Med.
1994;331:1689-1692.