Kruit, Mark Christian
Citation
Kruit, M. C. (2010, January 20). Migraine and brain lesions. Data from the population-based CAMERA Study. Department of Radiology, Faculty of Medicine, Leiden University Medical Center (LUMC), Leiden University.
Retrieved from https://hdl.handle.net/1887/14585
Version: Not Applicable (or Unknown) License:
Downloaded from: https://hdl.handle.net/1887/14585
Note: To cite this publication please use the final published version (if applicable).
C H A P T E R 2
EXISTINGEVIDENCEFORALINKBETWEENMIGRAINE
ANDWHITEMATTERLESIONS,SILENTINFARCTIONS,
ANDISCHEMICSTROKE
MarkC.Kruit
DepartmentofRadiology,LUMC
LenoreJ.Launer
LaboratoryofEpidemiology,DemographyandBiometry,
NationalInstituteonAging,NIH,Bethesda
MarkA.vanBuchem
DepartmentofRadiology,LUMC
GiselaM.Terwindt
DepartmentofNeurology,LUMC
MichelD.Ferrari
DepartmentofNeurology,LUMC
ADAPTEDFROM
HeadacheCurrents2005;2:6272
RevNeurol(Paris)2005;161:661665
From clinical practice we know that
features of migraine can occasionally
mimic stroke. Symptoms like visual dis
turbances, and less frequently somato
sensory symptoms, aphasia and hemi
paresis occur during MA but can also be
caused by cerebral ischemia. Typically,
aurasymptomsareslowinonset,slowin
progression, follow predictive patterns
(e.g. from occipital cortex to contiguous
temporalorparietalregions),andneuro
logical deficits cross arterial territories.
But presentation of symptoms may be
confusing,anditissometimesdifficultto
correctly distinguish migraine aura from
stroke or a transient ischemic attack
(TIA). This is notably the case when aura
symptoms remain longer than one hour
(or few hours in case of more than one
typeofauraatthesametime,e.g.visual,
somatosensory and motor symptoms).
When neurological symptoms remain
longer, up to 1 week, and results of
neuroimaging are normal, a diagnosis of
‘migraine with prolonged aura’ can be
given.† In case neuroimaging reveals (a)
relevantischemiclesion(s),adiagnosisof
‘migrainous infarction’ must be given,
regardlessofdurationofsymptoms.
† According to the new International Headache
Societyclassification,migrainewithprolongedaura
has been abandoned and subsequently classified
differently either as one of the subforms of 1.2
migrainewithaura,or1.6.2probablemigrainewith
aura.3
These situations illustrate that (sporadi
cally) after migraine attacks permanent
neurological deficits remain. Such dra
matic cases have been studied for many
years case wise, but the occurrence of
strokeinmigrainepatients,alsoirrespec
tive of a preceding migraine attack, has
alsoleadtoanumberof(hospitalbased)
casecontrol studies. The following para
graphsdetailonthecomplexrelationship
between migraine and clinical ischemic
stroke,migraineandsilentinfarctionand
migraineandMRIvisibleWMLs.
CLINICALISCHEMICSTROKE
INMIGRAINE
The relationship between migraine and
ischemicstrokeiscomplex.Thecomplex
ity is enhanced because of the wide
spreadconcurrentuseoforalcontracep
tives (OC), estrogen replacement ther
apy, vasoactive migraine medications,
and the presence of (other) concurrent
ischemicstrokeriskfactors.Anumberof
review articles provide an overview of
theproblem.7376Belowwetrytoamplify
shortlytherespectivecomponentsofthe
intriguing and important issue of clinical
ischemicstrokeinmigraine.
First, migraine can act as a direct
cause of ischemic stroke (migrainous
infarction). According to IHS criteria, the
diagnosis ‘migrainouscerebralinfarction’
can be applied in known MA patients,
when in a typical attack, one or more
aura symptoms are not fully reversible
within 7 days, or when there are neuroi
maging findings consistent with ischemic
stroke,andothercausesofinfarctionare
ruled out.2 It was suggested that a pro
longation of the migrainous process
beyond usual limits may explain most
migrainous infarctions.46 The annual
incidence of migrainous infarction was
estimated in a few studies, and ranged
between 1.43.4/100.000.45;73;77 Migrain
ousinfarctionisestimatedtoaccountfor
up to 40% of ischemic strokes in women
with migraine.62 But, because the strict
definition for migrainous infarction has
been inconsistently applied in several
studies, this rare condition is likely to be
overdiagnosed.76
Second, migraine may be sympto
matic or comorbid with other disorders
that may cause stroke. These mostly
vascular conditions, including CADASIL,78 cerebral arteriovenous malformations,
blood coagulation disorders,79 cardiac
abnormalities80;81 or mitochondrial dys
function (MELAS),82 can cause episodic
symptoms consistent with migraine
(usually with aura) but are also known
causes of ischemic stroke, and stroke
thenmayoccurremotelyfromamigraine
attack in these patients.83 Similarly,
stroke and migraine may just coexist
(remotely) in an otherwise healthy indi
vidual, without migraine being a contrib
utivefactortostroke.
Third, it has been suggested that
borderline perfusion of brain areas can
lower the threshold for developing mi
graine aura with or without headache.
Also around ischemic infarcts a zone of
reduced flow usually occurs, from which
cortical spreading depression may de
velop, that can simulate a migraine/aura
attack.84Inthisway,initialcerebralhypo
perfusion/ischemia, sometimes with
persisting deficits (stroke), may induce a
cooccurring migraine attack. This con
cept is illustrated by studies showing
increasedmigraineprevalenceinsubjects
with severe carotid stenosis related to
cervical artery dissection.85 Recent evi
denceshowsthat,onceinitiated,cortical
spreading depression can further con
tribute to the evolution of infarcts or to
an exacerbation of local cellular injury
causedbyischemia,byopeningtheBBB.
BBB disruption has been shown to con
tributetocelldeathandtoanincreasein
infarctvolumeincompromisedbrain.19;25 Fourth,migrainecanactasaninde
pendent risk factor for (clinical) ischemic
stroke. Several hospitalbased studies
comparedpatients with clinicalstroke to
controls without stroke with respect to
migraine diagnosis.5164 A metaanalysis
based on 14 studies summarized the
evidence.65 The pooled relative risk (RR)
of ischemic stroke among patients with
any type of migraine headache was 2.2
(95%confidenceinterval[CI]:1.92.5).Six
studies provided data specified by mi
graine subtype, resulting in a pooled RR
of 2.3 (95% CI: 1.63.2) for those with
MA, and 1.8 (95% CI: 1.13.2) for those
with MO, compared to controls. These
datadidnotchangeinstratifiedanalyses
byage,butitneedstobeconsideredthat
a number of the pooled studies only
included younger (<45 years of age)
women.5153;58;59;62;64
For this younger
subgroup of female migraine patients,
the recent casecontrol studies of mi
graine and stroke show a consistent,
significant,andhomogeneousincreasein
the risk of about 3 times compared to
controls of the same age.5153;58;59;62
Results of these studies should be inter
pretedwiththemethodologicalproblems
describedbelow.
STUDYRESULTS
In the mentioned studies, female MA
patientsseemedtobeathigherrisk(RRs
ranging 3.88.4)51;58;59;62
than MO, who
showed a significantly increased risk
(oddsratio [OR]:3.0 [95%CI: 1.55.8]) in
only one study.58 Significantly increased
ORs have further been found in female
migrainepatientswithafamilyhistoryof
migraine.62 One study investigated the
effectofduration,frequencyandrecency
of migraine on the risk of ischemic
stroke.51 They found higher risk in those
with migraine with more than 12 years
duration, those with initial MA, and in
thosewithmorethan12attacksperyear
(OR: 10.4 [95% CI: 2.249]). Further, a
relation between increased frequency of
MAduringthemonthsprecedingischem
icstrokewasobserved.
In women with migraine below age
45 who smoked, the risk of ischemic
stroke was found to be higher than the
productofmigraineassociatedrisktimes
smokingassociated risk (OR for those
smoking migraine patients ranging 7.4
10.2). Among current heavy smokers,
migraineincreasedtheriskbyafactorof
3.4.58;62 Similarly, the coexistence of
migraineandahighbloodpressurehada
greater than multiplicative effect in one
study.62 In three studies the effect of OC
usewasevaluated,54;58;62summarizedina
pooled RR of 8.7 (95% CI: 5.115.1)
among female migraineurs using OCs.65 ORs were lower in those using low dose
(<50gestrogen)OCs,thaninthoseusing
higherdoseOCs.Theserelationshipsand
probable interactions are complex, and
the available data is relatively limited.
Further study in large samples is needed
toexploretheeffectsbetter.
Fewstructureddataexistontheto
pography of infarcts in stroke patients
with migraine. In a large series of 3500
patients with acute stroke, 130 (3.7%)
had active migraine, and 66 of these
were younger than 45 years. In the
younger patients, posterior circulation
involvement (55%) was characteristic.86 Also other casereports and small series
suggested an overrepresentation of
clinicalstrokeintheoccipitallobeand/or
the posterior cerebral artery territory in
migrainepatients.87
In summary, data from observa
tional studies suggested that migraine is
an independent risk factor for stroke,
notably in younger women. Those with
higher attack frequency and those with
MAseemtobeathigherrisk.Thecoexis
tenceofmigraineandOCuse,highblood
pressure, or smoking seems to increase
the risk. The combined effect of risk
factors must be investigated further.
Giventheverylowabsoluteriskofstroke
in young women (estimated at
5.5/100.000 annually),88 there is no
absolute contraindication to OC use in
young female migraine patients; how
ever, patients (as everyone) should be
recommendedtoquitsmokingandtouse
lowestrogencontent pills or progesto
gens only, particularly in MA.76 The con
sistency of casecontrol findings from
several countries and supporting evi
dence from prospective studies suggest
that the association is not an artifact of
study design or execution. However,due
to methodological limitations (see be
low), none of the studies mentioned
abovecanbeconsidereddefiniteproofof
the association between migraine and
(clinical)ischemicstroke.75
SILENTINFARCTION
Silent infarcts are defined by the pres
ence of a brain parenchymal defect of
vascularorigin(confirmedbyCTorMRI),
in the absence of a history of clinical
stroke or TIA, and in the absence of
neurological symptoms and signs. Old
infarctsseenonCTarecommoninacute
stroke patients, who have no history of
clinical stroke.89 Silent brain infarcts are
frequentlyseenonMRIinhealthyelderly
people, and are associated with an in
creased risk of dementia and a steeper
decline in cognitive function.90 Further
more,thosewithsilentbraininfarctsand
white matter lesions are at much in
creasedriskofstroke,whichcouldnotbe
explained by the major stroke risk fac
tors.91;92Little is known about the preva
lence of silent infarcts in the general
population.Inthoseaged55to70years,
the prevalence ranges from 1115%;93;94 andincreaseswithage,from8%inthose
aged55to59years,to23%inthoseaged
65to72years.94Toourknowledge,there
arenopopulationbasedprevalencedata
available on silent cerebral infarcts in a
population aged below 5055 years.
Consequently, we don’t know whether
the presence of silent infarcts in the
younger population also implies in
creased risk of future stroke and/or
cognitive decline. With respect to mi
graine, there are no previous studies
specifically assessing the prevalence of
silentinfarcts.
WHITEMATTERLESIONS
GENERALFACTS
WMLs are a prevalent finding in the
brains of healthy elderly subjects, are
alsonotuncommoninsubjectsbelow50
years of age. WMLprevalence increases
withincreasingage,andisrelatedtothe
presence of hypertension, hypercholes
terolemia and other cardiovascular risk
factors.95;96 Larger volumes of WMLs are
associated with silent infarction, and
impaired cognitive and physical func
tion.90;97 A recent study demonstrated
thattheincreasedriskofstrokewithhigh
grades of white matter lesions is inde
pendent of traditional stroke risk factors
and silent infarcts.98 Therefore authors
suggested that assessment of white
matterdiseasemaybevaluableinassess
ing future risk of stroke. Damage to the
white matter is hypothesized to be the
result of ischemic complications of vari
ous microvascular processes, such as
(even brief) ischemia,99 hypoglycemia,
energydeprivation,oxidativestress,100or
platelethyperaggregability.101
WMLSINMIGRAINEPATIENTS
From themoment MRI became available
for the evaluation of the brains of mi
graine patients, reports exist on the
presence of white matter hyperintense
signal abnormalities on T2weighted MRI
sequences in these cases.102;103 Numer
ous studies have reported on WMLs in
migraine patients but results are incon
sistentand/orconflicting.6670;104109
Some
only reported prevalence of WMLs in
migraine patients, without the use of a
control group;107109 most reported an
increasedprevalenceinmigrainepatients
compared to controls,6670 and a few did
not find a statistically significant differ
ence.104106 Some found no difference in
prevalence between MO and
MA,67;69;106;108
but two studies found a
higher prevalence in MA.104;109 Only one
studyfoundanincreasedprevalencewith
increasingmigraineattackfrequency,but
notethattheotherstudiesdidnotassess
the influence of attack frequency as an
indicatorofmigraineseverity.109
VariabilityindetectionrateofWMLs
maybeduetodifferencesinpopulations
with respect to clinical factors. In addi
tion, diagnostic criteria for white matter
diseasecanvarymarkedly,andwithouta
clear definition of diagnostic criteria, for
instance, dilated perivascular CSF spaces
could have been misclassified as lesions.
None of the previous studies used fluid
attenuated inversion recovery (FLAIR)
sequences to eliminate the latter prob
lem. Only one study differentiated be
tween deep and periventricular WMLs,69 and found only the deep white matter
affected;mostotherseriesjustdescribed
“multiple small focal areas of increased
T2 signal in the white matter”, without
indicationofsiteortotallesionvolume.
Despite the various limitations and
discrepancies in the previous MRI stud
ies, a metaanalysis based on the results
of seven casecontrol studies6670;105;106
was published.110 The authors concluded
thatsubjectswithmigraine(comparedto
controls)areathigherrisk(OR:3.9[95%
CI: 2.36.7]) for having WMLs, regardless
of comorbidities (such as cardiovascular
risk factors, demyelinating disease, in
flammatoryconditionsandvalvularheart
disease). The increased risk was also
found in younger subjects without co
occurring cerebrovascular disease risk
factors.Noneofthesestudieswaspopu
lationbased, so much of the reported
dataisprobablybasedonamoresevere
subgroupofmigrainepatients.
METHODOLOGICALISSUES
Theinterpretationoftheresultsofprevi
ousstudiesthatassessedtheriskofbrain
lesionsinmigrainepopulations,aswellas
those that assessed migraine as a risk
factor in strokepopulations, was ham
pered by several types of potential bias.
Suchsourcesofbiasaresummarized.
First, in both types of studies, con
flictingresultsmadeinterpretationofthe
results of all studies difficult. Second,
concomitant (cardiovascular) risk factors
werenottakenintoaccountproperlyina
majority of the previous work, leaving
potential associations uncorrected for
important confounders. Third, the ob
served associations are likely to be re
strictedtocertainsubpopulationsorage
groupsofpatients,andcannotbegener
alizedtothegeneralpopulation.Further,
not all studies used IHS criteria for mi
graine diagnosis, and oftenthere wasno
standard diagnosis of migraine subtype,
thusallowingforthepotentialofmisclas
sification.
In a number of studies migraine
prevalence was compared between
patients with (clinical) stroke and those
without stroke. In these hospitalbased
strokecasecontrolstudies,migrainewas
almost always diagnosed retrospectively,
which makes this procedure vulnerable
to misclassification and recallbias. In
these clinical casecontrol studies, refer
ralbias might have played a role when
migraine patients with symptoms of
ischemic stroke have been preferentially
referred to centers specialized in mi
graineorstrokeandcasesofstrokewith
migraine are then more likely to be
detected.Insuchstudies,adifferenttype
of classificationbias might occur with
respect to the diagnosis of a qualifying
ischemic event. For instance, TIAs might
bedifficulttodiscriminatefrommigraine
aura,particularlyincaseswithprolonged
aura.111 Migraine might then both be
overestimated (aura misclassified as
ischemia) and underestimated (real
ischemia misclassified as aura) as a risk
factor for TIAs or stroke. To overcome
this classification problem, neuroimaging
proofconsistentwithcerebralischemiais
mandatory in such studies. In our opin
ion, a standard brain CT scan used in
previous hospitalbased casecontrol
studies,isnotsuitablefordetectingacute
cerebralischemia,andonlyhasaplacein
excludinghemorrhagicstrokes.
Whereas even in cases with clinical
symptomsofstrokeorTIAthereisaneed
for an objective neuroimaging proof of
brain ischemia, accurate imaging of the
brain is obviously the only way to be
informed about the presence and extent
of subclinical brain lesions in migraine
patients.Prevalenceofsubclinicallesions
in migraine patients, mostly WMLs, has
been assessed in a number of clinic
based MRI casecontrol studies. In addi
tion to the inherent potential types of
bias and other shortcomings described
above,themostimportantproblemwith
these MRI studies is the presence of
selectionbias: all studies included mi
graine patients that were probably al
ready known to a migrainespecialized
neurologist. In this way, it is likely that
only the more severe migraine cases
were included. Results of those studies
can therefore not be generalized to a
common (populationbased) migraine
population. Studies on the prevalence of
subclinical infarcts in populationbased
migrainecasesarelacking.