ORIGINAL ARTICLE

Evidence for a Vascular Factor in Migraine

Mohammad S. Asghar, MD,1 Adam E. Hansen, PhD,2,3 Faisal M. Amin, MD,1

R. J. van der Geest, MS,4 Patrick van der Koning, MS,4 Henrik B. W. Larsson, MD, DMS,2,5

Jes Olesen, MD, DMS,1 and Messoud Ashina, MD, PhD, DMS1

Objective: It has been suggested that migraine is caused by neural dysfunction without involvement ofvasodilatation. Because dismissal of vascular mechanisms seemed premature, we examined diameter of extra- andintracranial vessels in migraine without aura patients.Methods: A novel high-resolution direct magnetic resonance angiography imaging technique was used to measurearterial circumference of the extracranial middle meningeal artery (MMA) and the intracranial middle cerebral artery(MCA). Data were obtained at baseline, during migraine attack, and after treatment with the migraine abortive drugsumatriptan (a 5-hydroxytryptamine agonist).Results: We found dilatation of both MMA and MCA during migraine attack (p ¼ 0.001). Sumatriptan administrationcaused amelioration of headache (p < 0.001) and contraction of MMA (p < 0.001), but MCA remained unchanged (p¼ 0.16). Exploratory analysis revealed that in migraine attacks with half-sided headache, there was only dilatation onthe headache side of MMA of 12.49% (95% confidence interval [CI], 4.16–20.83%) and of MCA of 12.88% (95% CI,3.49–22.27%) and no dilatation on the nonheadache side of MMA (95% CI, �4.27 to 11.53%) and MCA (95% CI,�6.7 to 14.28%). In double-sided headache we found bilateral vasodilatation of both MMA and MCA (p < 0.001).Interpretation: These data show that migraine without aura is associated with dilatation of extra- and intracerebralarteries and that the headache location is associated with the location of the vasodilatation. Furthermore,contraction of extracerebral and not intracerebral arteries is associated with amelioration of headache. Collectively,these data suggest that vasodilatation and perivascular release of vasoactive substances is an integral mechanism ofmigraine pathophysiology.

ANN NEUROL 2011;69:635–645

Migraine is the most prevalent neurological disorder,

with >80 million sufferers in Europe and the

United States.1 It is characterized by recurrent attacks of

severe headache that typically affect the patients in their

most productive years and thereby lead to absenteeism

from work, decreased working efficiency and causing a fi-

nancial loss of 27 billion per year in the European

Union.2 Despite recent advances, the mechanisms under-

lying migraine pain still remain elusive. For >100 years,

the role of vascular and neural mechanisms has been a

topic of intense debate.3 Supported by Wolff ’s long series

of impressive investigations,4 a vascular hypothesis of vas-

odilatation as causative for the migraine pain dominated

migraine research for most of the 20th century.5 More

recently, the neuronal hypothesis has, however, received

increasing support. Based on findings of brainstem acti-

vation,6,7 it was suggested that migraine pain is caused

by abnormal central interpretation of normal sensory

input in the trigeminal sensory system.8 Moreover, a cen-

tral locus of action of antimigraine drugs such as triptans

has been suggested.9 A recent study of nitroglycerin-

induced migraine using magnetic resonance angiography

(MRA) concluded that there was no evidence in the

study or in the literature for dilation of extra- or intrace-

rebral arteries during a migraine attack.10 The authors

requested more human studies to elucidate the issue.11

Convinced that vascular nociception is involved in mi-

graine, we applied a high-resolution 3T MRA to examine

View this article online at wileyonlinelibrary.com. DOI: 10.1002/ana.22292

Received Jul 20, 2010, and in revised form Aug 28, 2010. Accepted for publication Oct 1, 2010.

Address correspondence to Dr Ashina, Danish Headache Center and Department of Neurology, Glostrup Hospital, University of Copenhagen, Faculty

of Health Sciences, Nordre Ringvej 57, DK-2600 Glostrup, Copenhagen, Denmark. E-mail: ashina@dadlnet.dk

From the 1Danish Headache Center and Department of Neurology, Glostrup Hospital, Faculty of Health Sciences, University of Copenhagen, Copenhagen,

Denmark; 2Functional Imaging Unit, Glostrup Hospital, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark; 3Department of

Radiology, Glostrup Hospital, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark; 4Division of Image Processing, Department of

Radiology, Leiden University Medical Center, Leiden, the Netherlands; and 5Department of Clinical Physiology and Nuclear Medicine, Glostrup Hospital,

Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark.

Additional supporting information can be found in the online version of this article.

VC 2011 American Neurological Association 635

whether dilatation of extra- and/or intracerebral arteries

occurs during migraine. We induced a migraine without

aura attack by infusion of calcitonin gene-related peptide

(CGRP). CGRP is a strong vasodilator12 present in peri-

vascular nerve fibers13 and known to induce migraine

attacks indistinguishable from spontaneous attacks in mi-

graine suffers.14 The latter is supported by the fact that

triggering of migraine is itself a fundamental feature of

the disorder, so that a so-called spontaneous attack, trig-

gered for example by missed sleep or food, is not differ-

entiated in clinical practice. Our hypothesis was that

both intra- and extracerebral arteries would be dilated

during a migraine attack, and in half-sided attacks we

expected to see dilatation on the headache side. In addi-

tion, we examined the effect of the selective antimigraine

drug sumatriptan, a 5-hydroxytryptamine (5-HT1B/D)

agonist, on arterial circumference during migraine

attacks. We hypothesized that sumatriptan would con-

strict arteries in association with pain relief.

Patients and Methods

Experimental DesignTwenty-four patients with migraine without aura received intra-

venous infusion of 1.5lg/min CGRP over 20 minutes (for

inclusion and exclusion criteria, see Supplementary Material).

CGRP infusion induces headache in healthy volunteers,15 and

delayed migraine attacks in patients with migraine without

aura.14 The patients were instructed that during the CGRP

infusion they might develop headache and several hours later a

delayed headache that could have migrainelike features. It was

emphasized that not all develop migraine headache after CGRP

infusion and that completion of the study did not depend on

the development of headache. If a subject developed migraine

she/he would be treated with a 6mg subcutaneous injection of

sumatriptan.

Before start of infusion, we recorded the circumference of

the middle meningeal artery (MMA) and middle cerebral artery

(MCA), blood pressure, pulse, and end tidal CO2. Headache

intensity was recorded on a 0 to 10 verbal rating scale (VRS).

Adverse events (AEs) such as nausea and other sensations relat-

ing to drug side effects were also recorded. After 40 minutes,

the subjects were removed from the magnet, and every 30

minutes headache score (VRS) and AEs were recorded. If sub-

jects developed delayed headache meeting the migraine criteria

(see Migraine Facts Box), they were placed in the magnet again

and new MRA recordings of MMA and MCA circumference

were preformed. The headache was subsequently treated with a

subcutaneous injection of 6mg sumatriptan. Fifteen minutes

later, when sumatriptan had reached its maximal serum concen-

tration, MRA recordings of MMA and MCA circumference

were repeated. If the subjects did not develop delayed headache

or migraine after 7 to 8 hours, MRA recordings of MMA and

MCA were repeated before discharge.

Migraine Facts

Migraine is the most common neurological disorder,

affecting 10% to 15% of the adult population in the

Western world. Prevalence is highest at age 25 to 55

years.1 After dementia and possibly stroke, migraine is

the most costly neurological disorder in the European

community, costing more than €27 billion a year in

direct and indirect costs.2

The diagnostic criteria for migraine without aura

according to the second edition of the International

Headache Classification41:

At least 5 attacks fulfilling the following criteria:

• Headache attacks lasting 4 to 72 hours (untreated or

successfully treated).

Headache has at least two of the following

characteristics:

• Unilateral location.

• Pulsating quality.

• Moderate or severe pain intensity.

• Aggravation by or causing avoidance of routine

physical activity (eg, walking or climbing stairs).

During headache at least 1 of the following:

• Nausea and/or vomiting.

• Photophobia and phonophobia.

• Symptoms not associated with another disorder.

Data Acquisition and Imaging ProtocolsAll subjects reported to the laboratory headache free for at least

5 days. Coffee, tea, cocoa, other methylxanthine-containing

foods or beverages, and tobacco were not allowed for at least

12 hours before start of the study. The subjects were placed in

the supine position in the scanner room, and a venous catheter

(Venflon) was inserted into the left antecubital vein for infu-

sion. Blood samples were collected to determine the baseline

hematocrit value and potassium and sodium levels. The subjects

were monitored for blood oxygen saturation, blood pressure,

and heart rate continuously during the study using a Veris

monitor (Medrad, Warrendale, PA). End-tidal CO2 was moni-

tored by a capnograph (Datex, Helsinki, Finland).

MRAMRA was performed on a 3.0T Philips Achieva Scanner (Phi-

lips Medical Systems, Best, the Netherlands) using an 8-element

phased-array receive head coil.

Using a high-resolution MRA technique, we obtained 2-

dimensional (2D) and 3D images of the extracranial MMA and

the intracranial MCA at baseline, during migraine attacks, and

ANNALS of Neurology

636 Volume 69, No. 4

after administration of sumatriptan. For vessel imaging, a 3D

inflow gradient echo sequence was used. First a scout MRA was

performed using field of view, 180 � 180 � 180mm3; acquired

matrix size (M � P), 180 � 180mm; acquired voxel resolution,

1.0 � 1.0 � 2.4mm3; reconstructed resolution, 0.34 � 0.34 �1.20mm3; repetition time, 23 milliseconds; echo time, 5.4

milliseconds; flip angle, 20�; SENSE factor, 2; 5 chunks, and

total scan duration, 3 minutes and 18 seconds. The scout MRA

was used to plan the subsequent MRA of MCA. MRA of

MCA was performed using field of view, 200 � 200 � 74mm;

acquired matrix size (M � P), 800 � 406; acquired voxel reso-

lution, 0.25 � 0.49 � 1.00mm3; reconstructed resolution, 0.20

� 0.20 � 0.5mm3; repetition time, 25 milliseconds; echo time,

3.5 milliseconds; flip angle, 20�; SENSE factor, 2; 4 chunks;

and total scan time, 9 minutes and 2 second. Angiography of

MMA was done using field of view, 200 � 200 � 16.1mm3;

acquired voxel resolution, 0.25 � 0.35 � 0.70mm3; reconstructed

resolution, 0.2 � 0.2 � 0.35mm3; repetition time, 25 milli-

seconds; echo time, 3.5 milliseconds; flip angle, 20�; SENSE factor,

2; 4 chunks; and total scan duration, 5 minutes and 28 seconds.

MRA Data AnalysisThe angiography data were transferred to a remote worksta-

tion in Dicom format. The MRA data where then analyzed

using the LKEB MRA Vessel wall analysis software from Lei-

den, the Netherlands.16 The reliability of measurements has

already been described and proven to be accurate with high

reproducibility by de Koning et al16 and applied in 2 previous

studies.10,17 The software provides automated contour detec-

tion and quantification of the luminal boundaries in vessel

segments with a minimal user interaction. In the 3D recon-

struction of the MRA data, the MCA was identified marking

the branch from the main trunk of the internal carotid artery.

In each scan, the exact same starting and ending point was

thus identified on MCA on both the right and the left side.

The software automatically calculated a path line and meas-

ured the circumference and mean diameter (mean value between

the measured maximum and minimum diameter) of the vessel.

The software did measurements every 0.2mm along the length

of the vessel perpendicular to the centerline. Analysis of MMA

was done in a similar manner, where MMA was identified on

both the right and left side marking the branch from the main

trunk of the maxillary artery. The data were then transferred to

Microsoft Excel and the mean of the measured variables (MCA

and MMA circumference) was calculated.

Statistical AnalysisAll absolute values are presented as mean 6 standard deviation.

Percent changes are reported as mean and 95% confidence

interval (CI) or mean 6 standard error of the mean. Headache

score is presented as median and quartiles. Baseline recordings

were obtained before administration of CGRP and included

MRA measurements, headache score, adverse events, and re-

cording of vascular variables.

We first tested for side-to-side differences in MMA and

MCA circumference at baseline using paired t tests. Because

there was no statistically significant difference, we first used the

mean of the right and left sides in the further testing. MMA

and MCA circumference, respectively, were tested: (1) during

the delayed headache phase against baseline using paired sample

t-test and (2) after treatment with sumatriptan against delayed

headache recordings using paired sample t test.

In patients who reported half sided headache, we exam-

ined changes in MMA and MCA circumference on the head-

ache side and the nonheadache side separately. The headache

and nonheadache sides were compared using baseline corrected

values. In case of bilateral headache, we compared changes in

right and left side MMA and MCA separately. We also tested

for differences between the right and left side using the baseline

corrected values. Five percent (p < 0.05) was accepted as the

level of significance. Statistical analysis was done using SPSS for

Mac 18.0 (SPSS Inc., Chicago, IL).

Results

Twenty-four patients completed the study (20 women

and 4 men; mean age, 32 years [range, 21–54 years];

FIGURE 1: Twenty-four patients with migraine without auracompleted the study. All participants received anintravenous infusion of calcitonin gene-related peptide(CGRP) (1.5lg/min over 20 minutes). Five patients reportedno delayed headache or migrainelike attacks. One of 5withdrew due to unspecified reasons. Magnetic resonanceangiography (MRA) was performed in the remaining 4patients 7 to 8 hours after the start of infusion. Nineteen of24 patients reported delayed headache, and 18 patients(75%) fulfilled the criteria for migrainelike attacks. Datafrom 1 patient had to be removed due to movementartifacts. Two patients withdrew due to claustrophobia.MRA was performed in the remaining 15 patients (5patients [33.3%] reported bilateral head pain, and 10patients [66.7%] reported unilateral head pain). All 15patients received a subcutaneous 6mg injection (Inj.) ofsumatriptan. Three patients did not complete the last MRAscan (2 patients due to worsening of headache and severenausea; 1 patient due to exhaustion and phonophobia).

Asghar et al: Vascular Factor in Migraine

April 2011 637

mean weight, 66kg [range, 48–83kg]). The study profile

is shown in Figure 1. Figure 2A shows headache intensity

in 18 patients (75%) who reported delayed migrainelike

attacks. Headache characteristics of the usual spontaneous

migraine attacks and CGRP-induced delayed migraine

attacks in the 18 participants are shown in the Table.Sixteen patients received a subcutaneous 6mg injection ofsumatriptan, and all but 1 patient reported a significantreduction in pain intensity from a median of 4 to 0

(VRS) (p < 0.001). Figure 2B compares the medianheadache score during migraine attack and after treat-ment with sumatriptan.

Blood and Vital SignsBaseline blood samples showed normal hematocrit, potas-sium, and sodium levels. There were no significantchanges in blood pressure, heart rate, end-tidal CO2, and

respiration frequency during the study (p 0.05), althougha noticeable but insignificant slight rise in blood pressureafter sumatriptan administration was observed (see Sup-plementary Table and the statistical analysis paragraph in

the Supplementary Material).

MMA and MCA Circumference during MigraineAttacks and after SumatriptanFifteen patients completed MRA scans during delayed

attacks, and 12 of 15 patients completed MRA scans after

sumatriptan administration. Mean MMA circumference

increased significantly from 4.49 6 0.52mm at baseline to

4.89 6 0.60mm during migraine attacks (p ¼ 0.001)

(Fig 3A). MMA circumference 30 minutes after sumatrip-

tan, at 3.86 6 0.52mm, was significantly smaller than the

value at baseline, 4.94 6 0.60mm (p < 0.001) (see Fig 3B).

MCA circumference during migraine attacks, 10.48

6 1.79mm, was significantly larger than at baseline,

9.46 6 1.51mm (p ¼ 0.005) (see Fig 3A). There was no

difference in MCA circumference before (9.79 61.76mm) and 30 minutes after sumatriptan (10.34 62.0mm) (p ¼ 0.16) (see Fig 3B).

MMA and MCA Circumference Changesin Unilateral Migraine AttacksMRA scans of 10 patients who reported unilateral head

pain during delayed migraine attacks showed dilatation

of MMA on the pain side, 5.19 6 0.75mm, compared

to baseline, 4.62 6 0.44mm (p ¼ 0.008). There was no

difference in MMA circumference on the nonpain side,

4.79 6 0.77mm, compared to baseline, 4.62 6 0.59mm

(p ¼ 0.291). Exploratory analysis (baseline corrected val-

ues) revealed significant dilatation on the pain side, 0.57

6 0.53mm, compared to the nonpain side, 0.17 60.48mm (p ¼ 0.032) (see Figs 3C and 4A).

After treatment with sumatriptan, MMA circumfer-

ence was significantly reduced on both the pain (p ¼0.003) and non-pain (p ¼ 0.002) sides compared to pre-

treatment. After sumatriptan, there was no difference

between the pain and nonpain sides (p ¼ 0.251) (see

Figs 3D and 4A).

FIGURE 2: (A) Median headache intensity in 18 patients(out of 24) who reported migraine attacks (75%). Thepatients reported immediate headache with a maximummedian intensity of 2 on the 0 to 10 verbal rating scale(VRS) at 25 minutes after the start of calcitonin gene-relatedpeptide (CGRP) infusion. The maximum immediateheadache lasted 25 minutes. The median headache intensitygradually decreased in the next 3 hours. The medianheadache intensity then gradually increased to 4 (VRS) at5.5 hours after the start of infusion. Sixteen patientsreceived 6mg sumatriptan by subcutaneous injection andreported a significant reduction in pain intensity from amedian of 4 to 0 (VRS) (p < 0.001). The characteristics ofthe usual spontaneous migraine attacks and provokedattacks for all 18 patients are shown in the Table. (B)Median headache intensity (a box-and-whisker diagram) ona 0 to 10 VRS during CGRP infusion, during migraine attack,and after sumatriptan administration. At baseline, allpatients were headache free (VRS 5 0). During the CGRPinfusion phase, the median headache intensity was 2.5.Several hours later, when the patients experienced amigraine attack, the headache intensity was 5; sumatriptanadministration had a dramatic effect on the headache andreduced the median headache intensity to 0.

ANNALS of Neurology

638 Volume 69, No. 4

TABLE 1: Headache Characteristics for Usual Spontaneous Migraine Attack and Delayed Migraine Attack

Subject Time forDelayedHeadache, hr

HeadacheCharacteristicsa

AssociatedSymptoms

MimicsMigraineAttack

Treatment(time)/Efficacy

1

A Unilateral/pressure þ/�/�B 5 Unilateral/pressure/6 þ/�/� Yes Sumatriptan (5.5 hours)/yes

2

A Unilateral/pressure þ/þ/þB 7.5 Unilateral/pressure/7 þ/þ/þ Yes Sumatriptan (8 hours)/yes

3

A Unilateral/throbbing þ/�/�B 6.5 Unilateral/throbbing/6 þ/þ/þ Yes Sumatriptan (7 hours)/yes

4

A Bilateral/throbbing þ/þ/þB 10 Bilateral/throbbing/3 þ/þ/þ Yes Sumatriptan (7.5 hours)/yes

5

A Bilateral/throbbing þ/þ/þB 6.5 Bilateral/throbbing/3 þ/�/þ Yes Sumatriptan (7 hours)/yes

6

A Unilateral/pulsating �/�/þB 10 Unilateral/pulsating/4 �/�/þ Yes Sumatriptan (10.5 hours)/yes

7

A Unilateral/pressure þ/þ/þB 10 Unilateral/pressure/4.5 þ/þ/þ Yes Sumatriptan (10.5 hours)/yes

8

A Unilateral/pressure �/þ/þB 6.5 Unilateral/pressure/6.5 �/þ/þ Yes Sumatriptan (7 hours)/yes

9

A Bilateral/throbbing þ/þ/þB 5 Bilateral/throbbing/6 þ/þ/� Yes Sumatriptan (5.5 hours)/yes

10

A Bilateral/pressure þ/þ/�B 5.5 Bilateral/throbbing/4 þ/þ/� Yes Sumatriptan (6 hours)/yes

11

A Unilateral/throbbing þ/þ/þB 10.5 Bilateral/throbbing/5 þ/þ/þ Yes Sumatriptan (11 hours)/yes

12

A Unilateral/throbbing þ/�/�B 8 Unilateral/throbbing/8 þ/�/þ Yes Sumatriptan (8.5 hours)/yes

Asghar et al: Vascular Factor in Migraine

April 2011 639

There was significant dilatation of MCA in the

pain side during delayed migraines, 10.66 6 2.03mm,

compared to baseline, 9.49 6 1.62mm (p ¼ 0.015).

MCA circumference on the nonpain side during delayed

migraines, 10.57 6 2.01mm, was not different from

baseline, 10.27 6 1.91mm (p ¼ 0.595) (see Figs 3E and

4B). Exploratory analysis showed no significant difference

from the pain side (p ¼ 0.068). After treatment with

sumatriptan, MCA circumference on the pain side, 10.0

6 1.81mm, was not different from pretreatment, 10.61

6 2.47mm (p ¼ 0.388). The nonpain side, at 9.53 61.70mm, was not different from baseline at 10.28 6

2.40mm (p ¼ 0.243) (see Figs 3F and 4B). There was

no difference between the headache side, 0.36 6

0.96mm, and the nonheadache side, �0.84 6 1.06mm

(baseline corrected values) (p ¼ 0.092).

MMA and MCA Circumference Changesin Bilateral Delayed Migraine AttacksWe performed MRA scans in 5 patients who reported

bilateral head pain during delayed attacks. The right and

left sides were analyzed separately.

On the right side, there was significant dilatation of

MMA, 5.04 6 0.41mm, compared to baseline, 4.60 6

0.32mm (p ¼ 0.015). On the left, there was also signifi-

cant dilatation of MMA, 4.37 6 0.92mm, compared to

baseline, 3.84 6 0.76mm (p ¼ 0.008). We found no

difference (baseline corrected values) between the right

TABLE 1 (Continued)

Subject Time forDelayedHeadache, hr

HeadacheCharacteristicsa

AssociatedSymptoms

MimicsMigraineAttack

Treatment(time)/Efficacy

13

A Unilateral/pressure �/þ/þB 9 Unilateral/pressure/4 �/þ/þ Yes Sumatriptan (9 hours)/yes

14

A Unilateral/pulsating þ/þ/þB 6.5 Unilateral/pulsating/4 þ/þ/þ Yes Sumatriptan (7 hours)/yes

15

A Unilateral/throbbing þ/þ/þB 10 Unilateral/throbbing/5.5 þ/þ/þ Yes Sumatriptan (11 hours)/yes

16

A Unilateral/pulsating þ/�/�B 3.5 Unilateral/throbbing/10 þ/þ/þ Yes Sumatriptan (4 hours)/nosumatriptan

(6 hours)/noNSAID (7 hours)/ yes

17

A Bilateral/pressure �/�/þB 9 Bilateral/throbbing/5 þ/þ/þ Yes N/A

18

A Unilateral/pressure þ/þ/þB 8 Unilateral/pressure/6 þ/þ/þ Yes Sumatriptan (8 hours)/nosumatriptan

(9 hours)/yes

19

A Bilateral/pressure �/þ/þB 5.5 Bilateral/pressure/4 �/�/� No N/A

Headache characteristics (localization, quality, and intensity) on a 0–10 verbal rating scale and associated symptoms (nausea, photo-phobia, and phonophobia) of 19 patients who developed delayed headache.A ¼ usual spontaneous headache/migraine attack characteristics; B ¼ delayed headache/migraine attack characteristics; NSAID ¼nonsteroidal anti-inflammatory drug; N/A ¼ not applicable.

ANNALS of Neurology

640 Volume 69, No. 4

FIGURE 3: (A) Mean relative circumference changes (%) standard error of the mean (SEM) of the middle meningeal artery andmiddle cerebral artery at baseline and during migrane. MMA dilated 9.24% 2.22 (95% CI, 4.49% to 14.01%) while MCAdilated 11.22% 3.09 (95% VI, 4.6% to 17.84%) during migraine. (B) Mean relative circumference changes (%) 6 SEM of MMAand MCA before and after sumatriptan administration. After sumatriptan MMA contracted by 221.49% 6 2.2 (95% CI,226.22% to 216.66%). The MCA was contracted by 24.9% 6 2.9 (95% CI, 210.9% to 1.05%). (C) Mean relativecircumference changes (%) 6 SEM in MMA on the pain side and non-pain side between baseline and during migraine inpatients with half-sided headache during migraine attack. On the pain side MMA dilated 12.49% 6 3.69 (95% CI, 4.16% to20.83%). In the non-headache side there was a non-significant dilatation of MMA of 3.63% 6 3.49 (95% CI, 24.27% to11.53%). (D) Mean relative circumference changes (%) 6 SEM in MMA on the pain and non-pain side after treatment withsumatriptan in patients with half-sided headache during migraine attack. After sumatriptan MMA contracted 220.45% 6 3.29(95% CI, 228.50% to 212.40%) on the pain side and 219.06% 6 3.23 (95% CI, 226.97% to 211.16%) in the non-pain side.(E) Mean relative circumference changes (%) 6 SEM in MCA between baseline and migraine on the pain side and the non-painside in patients with half-sided headache during migraine attack. On the pain side MCA dilated 12.88% 6 4.15 (95% CI, 3.49%to 22.27%). On the non-pain side there was no significant dilatation 3.79% 6 4.64 (95% CI, 26.7% to 14.28%). (F) Meanrelative circumference changes (%) 6 SEM in MCA on the pain and non-pain side after treatment with sumatriptan in patientswith half-sided headache during migraine attack. On neither side there was statistical difference compared to pre-treatment(the pain side: 24.5% 6 4.53 (95% CI, 215.6% to 6.58%); the non-pain side: 26.07% 6 3.95 (95% CI, 215.72 to 3.59%).

Asghar et al: Vascular Factor in Migraine

April 2011 641

MMA, 0.43 6 0.23mm, and the left MMA, 0.52 6

0.24mm (p ¼ 0.392). After sumatriptan, there was sig-

nificant reduction of MMA circumference on both sides

(right side, p ¼ 0.004; left side, p ¼ 0.012). Exploratory

analysis revealed no difference between sides (p ¼0.119). (Supplementary Fig A).

There was significant dilatation of the right MCA

during delayed attacks, 9.99 6 2.02mm, compared to

baseline, 8.56 6 1.14mm (p ¼ 0.033). There was also

significant dilatation of the left MCA during attacks,

10.41 6 1.27mm, compared to baseline, 8.69 6 0.87mm

(p ¼ 0.021). There was no difference (baseline corrected

values) between the right MCA, 1.43 6 1.0mm, and the

left MCA, 1.73 6 1.05mm (p ¼ 0.671).

After sumatriptan, there was no reduction in MCA

circumference on either sides (right side, p ¼ 0.701; left

side, p ¼ 0.200) and no difference between the right and

left sides (p ¼ 0.057) (Supplementary Fig B).

MMA and MCA Circumference in PatientsReporting No Delayed AttacksMRA scans in 4 patients who did not report delayed head-

aches or migraine attacks were recorded 7 to 8 hours after

the start of CGRP infusion. MMA or MCA circumference

did not differ between postinfusion (MMA, 4.28 60.48mm; MCA, 10.76 6 1.57mm) and baseline (MMA,

4.3 6 0.41mm; MCA, 10.87 6 1.74mm) (p > 0.05).

Discussion

The major finding of the present study was that CGRP

induced delayed migraine attacks and that the migraine

attacks were associated with dilatation of both extra- and

intracranial vessels. Patients with unilateral migraine pain

had dilatation of the MMA and MCA on the pain side

but not on the pain-free side. In patients with bilateral mi-

graine pain, we recorded bilateral dilatation of both the

MMA and MCA. Moreover, sumatriptan reversed dilata-

tion of the MMA and aborted migraine attacks. We found

no effect of sumatriptan on the MCA circumference.

Scientists have long sought the origin of migraine

pain. In 1672, Thomas Willis proposed the first vascular

hypothesis of migraine. This was supported by Graham

and Wolff,18 who found extracranial arterial dilatation

during migraine attack. Blood vessel involvement in mi-

graine was further suggested by dilatation of pial arteries

and flow changes associated with experimental cortical

spreading depression19–21 and with migraine attacks with

aura.22–25 Ultrasonography studies during migraine

attacks without aura have shown a slight but significant

arterial dilatation.26,27 By itself, the dilatation was not

sufficient to physically cause head pain, but rather it was

thought to reflect leakage of dilator and pronociceptive

substances in the perivascular space or parasympathetic

efferent effects. These findings have been challenged by a

Doppler study28 and more recently in an MRA study.10

In contrast to the present data, the latter MRA study

reported that provoked migraine attacks are not associ-

ated with dilatation of the MMA and the MCA. In this

as well as in the present study the same magnetic reso-

nance imaging scanner (Philips 3T) and head coil were

used to obtain MRA images, and the same vessel-wall

software was used to analyze the arteries from the same

anatomical localizations (the extracranial part of the

MMA and the initial part of the MCA). Although differ-

ent migraine pharmacological triggers (glyceryl trinitrate

in the previous study) were used, patients in both studies

reported migraine attacks indistinguishable from sponta-

neous attacks. However, the major differences between

the 2 studies are that (1) we applied predefined pain in-

tensity criteria (VRS � 4) before a migraine attack

FIGURE 4: (A) Three-dimensional (3D) data of middlemeningeal artery (MMA) in 1 subject with half-sidedheadache during migraine attack. MMA dilates on theheadache side during migraine attack, whereas no dilatationis observed on the nonheadache side. After injection ofsumatriptan, MMA contracts on both sides. (B) 3D data ofmiddle cerebral artery (MCA) in 1 subject. MCA dilates onthe headache side during migraine attack, whereas nodilatation is observed on the nonheadache aside. Afterinjection of sumatriptan no contraction of MCA is observed.

ANNALS of Neurology

642 Volume 69, No. 4

qualified for MRA scans; (2) we calculated the artery cir-

cumference17 and not the average diameter, because

arteries do not always resemble a circle, and the intracra-

nial anatomy does not always permit uniform dilatation

of the arteries; and (3) to avoid possible partial volume

problems, and because the vessel wall analysis software

requires at least 3 voxels across the measured artery, we

used a high-resolution MRA with a very small voxel size

(MMA, 0.2 � 0.2 � 0.35mm and MCA, 0.2 � 0.2 �0.5mm), resulting in images with sufficient voxels across

the vessels (2.8 � 5 voxels across MMA and 2 � 5 vox-

els across MCA). In contrast, the previous study used an

image resolution with approximately double the voxel

size (MMA, 0.39 � 0.39 � 0.25mm and MCA, 0.43 �0.43 � 0.65mm), allowing only 2 � 2 voxels across

MMA and 1.5 � 2.3 voxels across MCA10 (for data ac-

quisition, imaging protocols, and MRA analysis methods,

see Supplementary Material). Thus, the previous study

might have been unable to detect the subvoxel size dilata-

tion of MMA and MCA observed in the present study.

At the present field strength (3T), the image resolution

we applied was the maximally practically feasible. Other-

wise, each scan would have been more time-consuming

and result in low compliance, because it is rather

unpleasant for patients to be in a scanner during a mi-

graine attack. For all these reasons, our finding of arterial

dilatation during a migraine attack is highly likely to be

correct.

These data thereby clearly demonstrate that extra-

and intracranial arteries are only dilated on the pain

side during half-sided migraine attacks, and that bilat-

eral dilatation is observed in double-sided headache

during migraine attacks. It is well known that CGRP

has a rather short half-life (�10 minutes),29 which

rules out the possibility that the dilatation observed can

be due to a carryover effect due to the initial CGRP

infusion. This is further supported by the present

study, in which the postinfusion control scan revealed

no dilatation in the patients who did not develop mi-

graine attacks.

The question arises whether the modest dilatation

by 9 to 12% reported here is sufficient to activate nor-

mal sensory afferents in the perivascular space or whether

it has an effect only on sensitized nociceptors. We suggest

the latter, because arterial diameters may dilate markedly,

for example, during blood pressure decreases, and pulsa-

tion may increase during physical exercise without

accompanying head pain. But what are the possible

mechanisms of sensitization and excitation of the nerve

terminals? If migraine attacks start in the brain30 and

cause nociception, this can only be caused by efferent ac-

tivity in perivascular nerves.31 Sensitization could be

caused by leakage of vasoactive/sensitizing substances

from trigeminal nerve terminals32,33 or by efferent activ-

ity in parasympathetic nerves34 and consequent release of

substances that may sensitize sensory afferents. We con-

clude that migraine attacks may be evoked in deep brain

structures, but these initiating mechanisms trigger peri-

vascular release of vasoactive substances causing sensitiza-

tion of trigeminal afferents, vasodilatation, and migraine

pain.

Is the Mode of Action of Sumatriptan Vascularor Neuronal?Sumatriptan (5-HT1B/1D receptor agonist) was originally

developed as a selective cranial vasoconstrictor.35 How-

ever, the mode of action of sumatriptan and other trip-

tans in migraine is far more complex than initially per-

ceived.36 A central site of action has received support in

recent years, and vascular mechanisms have been

doubted.37 In the present study, we found that subcuta-

neous sumatriptan reversed dilatation and even con-

tracted MMA but not MCA during migraine attacks.

Simultaneously, all patients reported amelioration of mi-

graine headache. In the absence of a placebo group, the

specificity of MMA response suggests that the reversal

response is attributable to sumatriptan. These data pro-

vide the strongest evidence to date that sumatriptan

blocks dilatation in the major artery supplying blood to

the dura mater during migraine attacks. Given that

sumatriptan significantly constricted the MMA and not

the MCA, the present data suggest that the vascular site

of action of sumatriptan is primarily extracranial. These

results are supported by a recent MRA study in healthy

subjects17 showing that sumatriptan induced vasocon-

striction of meningeal but not intracerebral artery and

reduced head pain intensity. Interestingly, previous stud-

ies have shown that sumatriptan also blocks plasma pro-

tein extravasation, vasodilatation, and activation of sen-

sory afferents in dura in the neurogenic inflammation

model of migraine.38 The neurogenic inflammation

model has been used as an assay to predict probable anti-

migraine efficacy without causing vasoconstriction. How-

ever, several highly effective compounds in animal mod-

els failed to relieve migraine in clinical trials (for review

see Williamson and Hargreaves39). In another animal

model of migraine, triptans exerted their antimigraine

action through presynaptic 5-HT1B/1D receptors in the

nucleus caudalis, thus blocking synaptic transmission

between axon terminals of the peripheral trigeminovascu-

lar neurons and cell bodies of their central counter-

parts.40 Collectively, the results from human and animal

studies suggest that sumatriptan exerts its antimigraine

Asghar et al: Vascular Factor in Migraine

April 2011 643

effect by a combined constriction of extracerebral arteries

and inhibition of nociceptive input from these arteries.

ConclusionsThe major outcome of the present study is that migraine

attacks are associated with extra- and intracerebral arterial

dilatation, in particular on the headache side. Further-

more, contraction of dural arteries by sumatriptan is

associated with headache relief, suggesting primarily an

extracerebral site of vascular action of sumatriptan. Cra-

nial blood vessels thus seem to be important players in

the pathophysiology of migraine.

Acknowledgment

The study was supported by grants from the University

of Copenhagen, the Danish Headache Society, the Lund-

beck Foundation through the Center for Neurovascular

Signaling, Danish Agency for Science, Technology, and

Innovation, and Danish Council for Independent

Research–Medical Sciences (grant 271-08-0446). The

study was also supported in part by a research grant

(33936) from the Investigator Initiated Studies Program

of Merck & Co., Inc. The opinions expressed in this pa-

per are those of the authors and do not necessarily repre-

sent those of Merck & Co., Inc.

The authors thank all participating subjects; Dr S. Pe-

dersen for his insights and advice during the study;

radiographers M. Lindhart, B. S. Møller, and H. Simon-

sen for their work during the experiments; and medical

students K. D. Berg, T. R. Kapijimpanga, V. Kristensen,

and C. Vorbeck for their assistance in the MRA scans

and for their work in analyzing the data.

Potential Conflicts of Interest

J.O. has received grants and/or research support from, has

been a consultant and/or scientific adviser for, and has been

in the speakers’ bureau of Allergan Inc, AstraZeneca

Pharmaceuticals LP, Boehringer Ingelheim, Eli Lilly,

GlaxoSmithKline, Janssen Pharmaceutical Products, Lund-

beck, Merck, and Pfizer. M.A. has received grant support

and honoraria for lecturing from Merck, and honoraria for

lecturing from Pfizer, GlaxoSmithKline, Norpharma, and

AstraZeneca, and is a consultant and/or scientific adviser

for Merck and BTG international. There are no conflicts of

interest for the remaining authors.

References1. Lipton RB, Bigal ME, Diamond M, et al. Migraine prevalence, dis-

ease burden, and the need for preventive therapy. Neurology2007;68:343–349.

2. Andlin-Sobocki P, Jonsson B, Wittchen H, Olesen J. Cost of disor-ders of the brain in Europe. Eur J Neurol 2005;12(suppl 1):1–27.

3. Isler H, Koehler PJ. History of the headache. In: Olesen J, Tfelt-Hansen P, Welch KM, et al. The headaches. 3rd ed. Philadelphia,PA: Lippincott Williams & Wilkins, 2005:1–7.

4. Ray BS, Wolff HG. Experimental studies on headache: pain-sensi-tive structures of the head and their significance in headache.Arch Surg 1940;41:813–856.

5. Brennan K, Charles A. An update on the blood vessel in migraine.Curr Opin Neurol 2010;23:266–274.

6. Weiller C, May A, Limmroth V, et al. Brain stem activation in spon-taneous human migraine attacks. Nat Med 1995;1:658–660.

7. Bahra A, Matharu MS, Buchel C, et al. Brainstem activation spe-cific to migraine headache. Lancet 2001;357:1016–1017.

8. May A, Goadsby PJ. The trigeminovascular system in humans:pathophysiologic implications for primary headache syndromes ofthe neural influences on the cerebral circulation. J Cereb BloodFlow Metab 1999;19:115–127.

9. Sakai Y, Dobson C, Diksic M, et al. Sumatriptan normalizes the mi-graine attack-related increase in brain serotonin synthesis. Neurol-ogy 2008;70:431–439.

10. Schoonman G, van der Grond J, Kortmann C, et al. Migraineheadache is not associated with cerebral or meningeal vasodilata-tion—a 3T magnetic resonance angiography study. Brain 2008;131(pt 8):2192–2200.

11. Schoonman GG, Ferrari MD. Reply to: Migraine headache is notassociated with cerebral or meningeal vasodilatation—a 3T mag-netic resonance angiography study. Brain 2008;132:e113.

12. Brain SD, Williams TJ, Tippins JR, et al. Calcitonin gene-relatedpeptide is a potent vasodilator. Nature 1985;313:54–56.

13. McCulloch J, Uddman R, Kingman TA, Edvinsson L. Calcitoningene-related peptide: functional role in cerebrovascular regula-tion. Proc Natl Acad Sci U S A 1986;83:5731–5735.

14. Lassen L, Haderslev P, Jacobsen V, et al. CGRP may play a causa-tive role in migraine. Cephalalgia 2002;22:54–61.

15. Petersen K, Lassen L, Birk S, et al. BIBNBS$6$7 antagonizes humanalpha-calcitonin gene related peptide-induced headache and extra-cerebral artery dilatation. Clin Pharmacol Ther 2005;77:202–213.

16. de Koning P, Schaap J, Janssen J, et al. Automated segmentationand analysis of vascular structures in magnetic resonance angio-graphic images. Magn Reson Med 2003;50:1189–1198.

17. Asghar MS, Hansen AE, Kapijimpanga TR, et al. Dilatation byCGRP of middle meningeal artery and reversal by sumatriptan innormal volunteers. Neurology 2010;75:1494–1495.

18. Graham JR, Wolff HG. Mechanism of migraine headache andaction or ergotamine tartrate. Arch Neurol Psychiatry 1938;39:737–763.

19. Wahl M, Schilling L, Parsons AA, Kaumann A. Involvement of calci-tonin gene-related peptide (CGRP) and nitric oxide (NO) in thepial artery dilatation elicited by cortical spreading depression.Brain Res 1994;637:204–210.

20. Lauritzen M, Fabricius M. Real time laser-Doppler perfusion imag-ing of cortical spreading depression in rat neocortex. Neuroreport1995;6:1271–1273.

21. Woods RP, Iacoboni M, Mazziotta JC. Brief report: bilateralspreading cerebral hypoperfusion during spontaneous migraineheadache. N Engl J Med 1994;331:1689–1692.

22. Olesen J, Larsen B, Lauritzen M. Focal hyperemia followed byspreading oligemia and impaired activation of rCBF in classic mi-graine. Ann Neurol 1981;9:344–352.

23. Lauritzen M, Olesen J. Regional cerebral blood flow during mi-graine attacks by xenon-133 inhalation and emission tomography.Brain 1984;107(pt 2):447–461.

ANNALS of Neurology

644 Volume 69, No. 4

24. Hadjikhani N, Sanchez Del Rio M, Wu O, et al. Mechanisms of mi-graine aura revealed by functional MRI in human visual cortex.Proc Natl Acad Sci U S A 2001;98:4687–4692.

25. Lauritzen M. Pathophysiology of the migraine aura: the spreadingdepression theory. Brain 1994;117:199–210.

26. Thomsen LL, Iversen HK, Olesen J. Cerebral blood flow velocitiesare reduced during attacks of unilateral migraine without aura.Cephalalgia 1995;15:109–116.

27. Iversen HK, Nielsen TH, Olesen J, Tfelt-Hansen P. Arterialresponses during migraine headache. Lancet 1990;336:837–839.

28. Zwetsloot CP, Caekebeke JF, Ferrari MD. Lack of asymmetry ofmiddle cerebral artery blood velocity in unilateral migraine. Stroke1993;24:1335–1338.

29. Brain S, Cambridge H. Calcitonin gene-related peptide: vasoactiveeffects and potential therapeutic role. Gen Pharmacol 1996;27:607–611.

30. Goadsby PJ, Charbit AR, Andreou AP, et al. Neurobiology of mi-graine. Neuroscience 2009;161:327–341.

31. Olesen J, Burstein R, Ashina M, Tfelt-Hansen P. Origin of pain inmigraine: evidence for peripheral sensitisation. Lancet Neurol2009;8:679–690.

32. Mayberg M, Langer RS, Zervas NT, Moskowitz MA. Perivascularmeningeal projections from cat trigeminal ganglia: possible path-way for vascular headaches in man. Science 1981;213:228–230.

33. Strassman AM, Raymond SA, Burstein R. Sensitization of menin-geal sensory neurons and the origin of headaches. Nature 1996;384:560–564.

34. Burstein R, Jakubowski M. Unitary hypothesis for multipletriggers of the pain and strain of migraine. J Comp Neurol 2005;493:9–14.

35. Humphrey PP, Goadsby PJ. The mode of action of sumatriptan isvascular? A debate. Cephalalgia 1994;14:401–410; discussion 393.

36. Ferrari MD, Goadsby PJ, Roon KI, Lipton RB. Triptans (serotonin,5-HT1B/1D agonists) in migraine: detailed results and methods ofa meta-analysis of 53 trials. Cephalalgia 2002;22:633–658.

37. Tfelt-Hansen PC. Does sumatriptan cross the blood-brain barrierin animals and man? J Headache Pain 2010;11:5–12.

38. Moskowitz MA. Neurogenic inflammation in the pathophysiology andtreatment of migraine. Neurology 1993;43(6 suppl 3):S16–S20.

39. Williamson DJ, Hargreaves RJ. Neurogenic inflammation in thecontext of migraine. Microsc Res Tech 2001;53:167–178.

40. Levy D, Jakubowski M, Burstein R. Disruption of communicationbetween peripheral and central trigeminovascular neurons medi-ates the antimigraine action of 5HT 1B/1D receptor agonists. ProcNatl Acad Sci U S A 2004;101:4274–4279.

41. Olesen J. The international classification of headache disorders.2nd edition (ICHD-II). Rev Neurol (Paris) 2005;161:689–691.

Asghar et al: Vascular Factor in Migraine

April 2011 645