Original Article

Triptans and serious adverse vascularevents: Data mining of the FDA AdverseEvent Reporting System database

Giuseppe Roberto1, Carlo Piccinni1, Roberto D’Alessandro2

and Elisabetta Poluzzi1

Abstract

Aim: The aim of this article is to investigate the vascular safety profile of triptans through an analysis of the United States

Food and Drug Administration Adverse Event Reporting System (FDA_AERS) database with a special focus on serious

and unexpected adverse events.

Methods: A case/non-case analysis was performed on the reports entered in the FDA_AERS from 2004 to 2010: Cases

were reports with at least one event included in the MedDRA system organ classes ‘Cardiac disorder’ or ‘Vascular

disorders’, whereas non-cases were all the remaining reports. Co-reported cardiovascular drugs were used as a proxy of

cardiovascular risk and the adjusted reporting odds ratio (adj.ROR) with 95% confidence intervals (95% CI) was

calculated. Disproportionality signals were defined as adj.ROR value >1. Adverse events were considered unexpected

if not mentioned on the relevant label.

Results: Among 2,131,688 reports, 7808 concerned triptans. Cases were 2593 among triptans and 665,940 for all other

drugs. Unexpected disproportionality signals were found in the following high-level terms of the MedDRA hierarchy:

‘Cerebrovascular and spinal necrosis and vascular insufficiency’ (103 triptan cases), ‘Aneurysms and dissections non-site

specific’ (15), ‘Pregnancy-associated hypertension’ (10), ‘Reproductive system necrosis and vascular insufficiency’ (3).

Discussion: Our analysis revealed three main groups of unexpected associations between triptans and serious vascular

events: ischaemic cerebrovascular events, aneurysms and artery dissections, and pregnancy-related vascular events.

A case-by-case assessment is needed to confirm or disprove their plausibility and large-scale analytical studies should

be planned for risk rate estimation. In the meantime, clinicians should pay special attention to migraine diagnosis and

vascular risk assessment before prescribing a triptan, also promptly reporting any unexpected event to pharmacovigi-

lance systems.

Keywords

Triptans, migraine, adverse events, vascular events, pharmacovigilance, adverse event reporting system, disproportion-

ality, data mining

Date received: 11 December 2012; revised: 28 March 2013; accepted: 28 May 2013

Background

Triptans are specific antimigraine drugs generally rec-ommended to patients in whom traditional analgaesicsor acetaminophen are ineffective (1). The therapeuticeffect of triptans appears to stem from their ability toactivate the serotoninergic 5HT1b/d receptors in the tri-gemino-vascular system, causing vasoconstriction ofthe cerebral vessels and neuronal inhibition. Despitetheir relative selectivity, the vasoconstrictive activityof triptans also affects coronary and peripheral arteries,although to a lesser extent (2,3). Most of the clinicalevidence on triptans comes from patients without

known vascular diseases (4). For these reasons triptansshould be avoided or prescribed with caution in sub-jects with a history of cardio- or cerebro-vascular

1Department of Medical and Surgical Sciences – Pharmacology Unit,

University of Bologna, Italy2IRCCS – Istituto delle Scienze Neurologiche di Bologna, Italy

Corresponding author:

Elisabetta Poluzzi, University of Bologna, Department of Medical and

Surgical Sciences – Pharmacology Unit, Via Irnerio, 48, I-40126 Bologna

BO, Italy.

Email: elisabetta.poluzzi@unibo.it

Cephalalgia

2014, Vol 34(1) 5–13

! International Headache Society 2013

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/0333102413499649

cep.sagepub.com

diseases and in those with pre-existing vascular riskfactors (5).

Although no epidemiological study to date hashighlighted an increased risk of serious vascular acci-dents (2), a number of published case reports havedocumented serious ischaemic adverse events after trip-tan use (6–9), and summaries of product characteristicsof triptan-containing medicines list ischaemic coronaryaccident (e.g. arteriospasm coronary, Prinzmetalangina, acute coronary syndrome, myocardial infarc-tion) and serious peripheral vascular events (e.g. ischae-mic colitis) among possible adverse reactions to triptanconsumption (5,10).

Overall, findings from clinical practice surveys haveshown that patients and clinicians seem to follow therecommendations for the safe use of triptans (11), butsome users concomitantly receive prescriptions thatsuggest a contraindication or a potentially increasedrisk of vascular accidents related to triptan intake (12).

Since triptan use in the general population is rela-tively low (a one-year prevalence around 1% withabout 50% of the incident users receiving only one pre-scription (12,13)) and serious ischaemic events appearto be very rare in patients eligible for triptan treatment(1:100,000 treated attacks with sumatriptan) (2), spon-taneous reporting data can add further information onthe vascular safety of triptans in clinical practice, par-ticularly on serious and infrequent adverse events. Toour knowledge, no independent studies have been pub-lished on this topic. Therefore, the aim of the presentanalysis was to investigate the vascular safety profile oftriptans by analysing the largest international spontan-eous reporting database (the United States (US) Foodand Drug Administration Adverse Event ReportingSystem (FDA_AERS)) with a special focus on seriousand unexpected adverse events.

Materials and methods

The FDA_AERS database collects spontaneous reportsof adverse events associated with the administration ofdrugs and therapeutic biologics, using specific submis-sion forms according to the MedWATCH programme(for details on required items, see http://www.fda.gov/downloads/Safety/MedWatch/HowToReport/DownloadForms/UCM082728.pdf). Reports aremainly submitted to the database by healthcare profes-sionals, consumers and manufacturers, from the USand other countries (14). Each report can includemore than one concomitantly occurring adverse event,and many administered drugs.

Since the first quarter of 2004, downloading of rawdata extracted from the AERS database has been madeavailable to the public from the FDA website (http://www.fda.gov/cder/aers/extract.htm).

Information included in each submitted report is rec-orded in seven different datasets: DEMO_file (patientdemographic and administrative information);DRUG_file (name of drugs/biologics and the relevantrole in the occurrence of adverse event; e.g. suspected);REAC_file (adverse events); OUTC_file (patient out-comes); RPSR_file (reporting sources); THER_file(date of therapy start/end) and INDI_file (indicationsfor use). Two codes, the ‘case number’ and the individ-ual safety report (ISR), link all the datasets. Moreover,in agreement with FDA_AERS recommendations (asdescribed in the README file included in each down-loadable dataset), possible duplicates of each singlereport were identified and merged by considering the‘case number’.

Adverse events are coded into the FDA_AERS usingthe preferred term (PT) level of the standardized ter-minology of the Medical Dictionary for RegulatoryActivities (MedDRA) version 13.0. According to thehierarchical structure of MedDRA, based on term spe-cificity, higher levels of the dictionary (HLT: high levelterms; HLGT: high-level group terms; SOC: systemorgan classes) are automatically assigned to each PT,providing a more general diagnostic classification of thecoded event (15).

For the purpose of the present analysis, all thereports submitted to the database FDA_AERS from2004 through 2010 were considered. Before performingstatistical analyses, the dataset was prepared mappingeach drug reported as suspected or interacting to itsgeneric name and to the relevant anatomical thera-peutic chemical (ATC) code (16–18).

In order to highlight those adverse events dispropor-tionally reported following the use of a triptan, a case/non-case analysis was performed and the reporting oddsratio (ROR) values, with the relevant 95% confidenceintervals (95% CI), were calculated by Mantel-Haenszel univariate analysis. The ROR is a dispropor-tional measure based on the ratio of the odds of cases intriptan reports in comparison with reports without trip-tans (Figure 1) (19). In the present analysis, adverseevents coded through PTs belonging to the MedDRASOCs ‘Cardiac disorder’ or ‘Vascular disorders’ wereconsidered outcomes of interest. Each report contain-ing one or more outcomes of interest was consideredone case, whereas non-cases were all the remainingreports other than cases. Disproportionality signalswere defined when cases were �3 and ROR was statis-tically >1 (i.e. lower limit of 95% CI >1).

Since the more general diagnostic classification pro-vided by higher levels of MedDRA can increase the sen-sitivity of signal detection (20), we performed theprimary analysis testing on all the HLGTs and HLTsbelonging to the SOCs considered. In contrast, the sec-ondary analysis concerned only PTs subordinated to

6 Cephalalgia 34(1)

those HLTs previously resulting in a disproportionalitysignal: The higher specificity of the PT level may betterdescribe the event associated with drug exposure (15,20).

As an additional analysis, six standardisedMedDRA queries (SMQs) considered suitable for thepurposes of the present study were also applied. SMQsare provided by the MedDRA dictionary to allowgrouping of terms describing the same medicalcondition (15).

Moreover, the presence in the report of drugslabelled to treat cardiovascular conditions (ATC codeC) was considered a rough proxy of possible pre-existing risk (see also Table 1S in the supplementarymaterial). This parameter was used to adjust rawROR values (adj.ROR) and to estimate the associationbetween drug and events independently from the pre-existing vascular risk.

The disproportionality signals obtained were con-sidered expected or unexpected on the basis of theinformation reported in the FDA PrescribingInformation of triptans (5) and the Summary ofProduct Characteristics of triptans issued by theMedicine and Healthcare products RegulatoryAgency (MHRA) (10).

Data management and analysis were performedusing the Microsoft Access Office 2003 and Epi Info,version 3.4.3 software packages.

Results

After data cleaning, the dataset included 2,131,688adverse event reports. Among these, 7808 included atleast one triptan as the suspected or interacting drug.The mean reported age of subjects exposed to triptanswas 44 years (total age range 0–90 years) with a female/male ratio of 4.8:1. Sumatriptan was the most fre-quently reported active ingredient of the class (73.1%of triptan reports), followed by eletriptan (10.8%, seeTable 1). Out of 668,533 reports in which at least one ofthe PTs listed was included in the ‘‘Cardiac disorder’’ or‘‘Vascular disorders’’ SOCs, 2593 concerned triptans.

In the primary analysis (see Figure 2 and supplemen-tary material) application of the ROR to all the HLGTs

included in the cardiac or vascular disorders SOCs(N¼ 21) yielded four signals of disproportionatereporting: ‘Coronary artery disorders’ (adj.ROR¼2.00; 95% CI¼ 1.85–2.16), ‘Cardiac disorders signand symptoms’ (1.11; 1.03–1.20) ‘Vascular disordersNEC (not elsewhere codified)’ (2.01; 1.34–2.99) and‘Aneurysms and artery dissections’ (2.01; 1.34–2.99).

Among 102 HLTs tested, ten resulted in a dispropor-tionality signal both before and after statistical adjust-ment for concomitant cardiovascular drugs (see Figure 2and supplementary material). The highest reporting dis-proportions suggested by the adj.ROR values were‘Ischaemic coronary artery disorders’ (adj.ROR¼ 2.08;95% CI¼ 1.93–2.25), ‘Cerebrovascular and spinal vas-cular disordersNEC’ (3.55; 3.27–3.87), ‘Gastrointestinalnecrosis and vascular insufficiency’ (2.82; 1.78–4.40),‘Aneurysms and dissections non-site specific’ (3.45;1.99–5.86) and ‘Vascular malformations and acquiredanomalies’ (2.36; 1.19–4.51). Two additional HLTsresulted in a signal only before the statistical adjustment:‘Ocular haemorrhagic disorders’ (ROR¼ 2.51; 95%CI¼ 1.48–4.25) and ‘Cerebrovascular aneurysm anddissections’ (1.64; 1.00–2.68).

In the secondary analysis, several triptan-PTcombinations showed significant adjusted ROR

Reports with event of interest

Reports without event of interest

Reports with triptans a b

Reports withouttriptans c d

c/d

a/bROR =

Figure 1. Calculation of ROR (reporting odds ratio): Contingency table and formula.

Table 1. Total reports and number of cases with at least one

adverse event in the MedDRA ‘‘Cardiac disorders’’ or ‘‘Vascular

disorders’’ SOCs per specific triptan (active substance).

Active

substance

Total reports

(n¼ 7,808)a %

Number of cases

(n¼ 2593)a %

Sumatriptan 5817 73.1 1659 62.6

Eletriptan 853 10.8 360 13.6

Zolmitriptan 507 6.4 237 8.9

Rizatriptan 354 4.5 201 7.6

Naratriptan 229 2.9 91 3.4

Frovatriptan 101 1.3 65 2.5

Almotriptan 66 0.8 39 1.5

aIn 101 cases more than one triptan was reported.

SOC: system organ classes.

Roberto et al. 7

(see Table 2): ‘Arteriospasm coronary’ (adj.ROR¼21.58; 16.06–28.86), ‘Prinzmetal angina’ (adj.ROR¼31.14; 95% CI¼ 16.87–56.94), ‘Cerebrovascularspasm’ (21.46; 10.13–43.63), ‘Cerebral vasoconstriction’(46.18; 17.39–114.04), ‘Artery dissection’ (12.24; 4.42–31.25) ‘Carotid artery dissection’ (27.21; 12.09–58.21),‘Raynaud’s phenomenon’ (6.44; 4.50–9.14) ‘Intestinalischaemia’ (4.35; 2.51–7.39) ‘Congenital arterial malfor-mation’ (41.12; 9.59–146.21) and ‘Placental infarction’(12.68; 3.23–42.70). The only statistically significantHLT that did not result in any signal in the secondaryanalysis was ‘Ocular haemorrhagic disorders’.

In the SMQ analysis, three out of six queries selectedfor the study resulted in a disproportionality signal (seeTable 3): ‘‘Ischaemic heart disease’’: (adj.ROR¼ 1.47;95% CI¼ 1.21–1.77), ‘‘Central nervous system haem-orrhages and cerebral conditions’’ (2.95; 1.54–5.47) and‘‘Cerebrovascular disorders’’ (2.25; 1.09–4.46).

Discussion

This is the first independent study to investigate thevascular safety profile of triptans in a large spontaneous

reporting database since the consensus statement oncardiovascular safety profile of triptans, which analysedthe FDA_AERS database up to 2002 (4). Our analysisdisclosed three main groups of unexpected associations:ischemic cerebrovascular events, aneurysms and arterydissections, and pregnancy-related vascular events. Allthe remaining associations highlighted in the study wereconsistent with the already known features of the trip-tan safety profile (5,10) (i.e. expected), thus confirmingthe validity both of the data-mining method appliedand the data source used.

Our method identified specific events that were morefrequently reported in association with the studieddrugs than with all the other drugs in the dataset. Inparticular, analyses of passive post-marketing surveil-lance data cannot estimate the incidence rate of anadverse event because of the lack of both the totalnumber of patients exposed to the drug of interestand the total number of patients experiencing a particu-lar adverse event (underreporting) (21). For this reason,our results should be interpreted with caution as a start-ing point for more specific studies (e.g. case-by-caseanalysis, case-control and cohort studies).

HLTsn. of analysed terms = 102

n. of disproportionality signals = 12

HLGTsn. of analysed terms = 21

n. of disproportionality signals = 4

SOCs

Cardiac disorder/Vascular disorder(N=2,593)

Coronary artery disorders (N=758) Ischaemic coronary artery disorders (N=750)

Cardiac disorder signs and symptoms (N=745) Cardiac signs and symptoms NEC† (N=478)

Vascular disorders NEC† (N=777)

Cerebrovascular and spinal vascular disorders NEC† (N=606)

Vascular malformations and acquired anomalies (N=10)

Aneurysms and artery dissections (N=26)

Aneurysms and dissections non site specific (N=15)-

Cerebrovascular aneurysms and dissections (N=16)

Arteriosclerosis, stenosis, vascularinsufficiency and necrosis (N=217)

Cerebrovascular and spinal necrosis and vascular insufficiency (N=103)

Peripheral vasoconstriction, necrosis and vascular insufficiency (N=50)

Gastrointestinal necrosis and vascular insufficiency (N=21)

Reproductive system necrosis and vascular insufficiency (N=3)

Vascular hypertensive disorders (N=124) Pregnancy associated hypertension (N=10)

Vascular haemorrhagic disorders (N=348) Ocular haemorrhagic disorders (N=14)

Figure 2. ‘Primary analysis’ of triptan-associated vascular events classified at different MedDRA levels: SOC: system organ class;

HLGT: high-level group term; HLT: high-level term. White boxes include terms resulting in disproportionality signals, while grey boxes

include terms without disproportionality; yNEC: not elsewhere codified; (crude and adjusted reporting odds ratios (RORs) for each

MedDRA term are reported in the supplementary material). MedDRA: Medical Dictionary for Regulatory Activities.

8 Cephalalgia 34(1)

Tab

le2.

Pre

ferr

ed

term

sre

sultin

gin

ast

atis

tica

llysi

gnifi

cant

report

ing

odds

ratio

(RO

R)

either

befo

reor

afte

rst

atis

tica

lad

just

ment

(seco

ndar

yan

alys

is).

Hig

h-leve

lte

rm(H

LT)

Pre

ferr

ed

term

s(P

T)a

Tri

pta

nca

ses

RO

RC

I95%

RO

Rb

(adju

sted)

CI

95%

Isch

aem

icco

rona

ryar

tery

diso

rder

sC

hest

pai

n48

2.3

22.0

7–2.6

02.5

32.2

6–2.8

5

Chest

dis

com

fort

40

5.9

65.2

9–6.7

26.3

65.6

4–7.1

8

Art

eri

osp

asm

coro

nar

y10

19.2

114.4

9–25.4

721.5

816.0

6–28.8

6

Pri

nzm

eta

lan

gina

525.3

014.3

3–4.6

831.1

416.8

7–56.9

4

Acu

teco

ronar

ysy

ndro

me

31.7

70.9

8–3.2

02.1

71.1

4–4.0

2

Cer

ebro

vasc

ular

and

spin

alva

scul

ardi

sord

ers

NEC

Cere

bro

vasc

ula

rdis

ord

er

13

2.4

31.4

1–4.2

02.5

11.3

9–4.4

2

Car

otid

arte

rydis

eas

e6

3.1

21.4

0–6.9

83.8

31.5

4–8.8

4

Car

diac

sign

san

dsy

mpto

ms

NEC

Pal

pitat

ions

139

2.1

11.7

9–2.5

02.2

91.9

3–2.7

2

Cer

ebro

vasc

ular

and

spin

alne

cros

is

and

vasc

ular

insu

ffic

ienc

y

Cere

bra

lin

farc

tion

26

1.5

01.0

2–2.2

11.7

11.1

4–2.5

4

Isch

aem

icst

roke

22

2.5

71.6

8–3.9

12.7

31.7

5–4.2

2

Cere

bro

vasc

ula

rsp

asm

922.6

911.4

9–44.8

121.4

610.1

3–43.6

3

Car

otid

arte

ryocc

lusi

on

72.1

21.0

1–4.4

52.6

01.1

4–5.6

5

Cere

bra

lva

soco

nst

rict

ion

640.8

317.3

1–96.3

446.1

817.3

9–114.0

4

Cere

bra

lar

tery

occ

lusi

on

53.9

21.6

2–9.4

84.4

11.6

1–11.0

1

Cere

bella

rin

farc

tion

52.7

51.1

4–6.6

44.3

41.5

8–10.8

4

Bra

inst

em

infa

rction

53.9

61.6

4–9.5

74.4

41.6

2–11.0

9

Periphe

ralva

soco

nstr

ictio

n,ne

cros

isan

dva

scul

arin

suffic

ienc

yR

aynau

d’s

phenom

enon

17

6.2

93.8

9–10.1

96.4

44.5

0–9.1

4

Gas

troi

ntes

tinal

necr

osis

and

vasc

ular

insu

ffic

ienc

yIn

test

inal

isch

aem

ia15

3.8

22.2

9–6.3

54.3

52.5

1–7.3

9

Ane

urys

ms

and

diss

ectio

nsno

n-site

spec

ific

Aneury

sm9

2.5

91.3

4–4.9

92.8

71.3

9–5.6

8

Art

ery

dis

sect

ion

512.3

75.0

5–30.3

212.2

44.4

2–31.2

5

Cer

ebro

vasc

ular

aneu

rysm

san

ddi

ssec

tions

Car

otid

arte

rydis

sect

ion

828.6

613.8

3–59.4

127.2

112.0

9–58.2

1

Preg

nanc

y-as

soci

ated

hyper

tens

ion

Pre

gnan

cy-induce

dhy

pert

ensi

on

57.5

23.0

9–18.2

97.6

12.7

5–19.1

4

Vasc

ular

mal

form

atio

nsan

dac

quired

anom

alie

sC

onge

nital

arte

rial

mal

form

atio

n3

42.9

712.7

1–145.2

241.1

29.5

9–146.2

1

Rep

rodu

ctive

syst

emne

cros

isan

dva

scul

arin

suffic

ienc

yPla

centa

lin

farc

tion

313.8

44.3

4–44.1

512.6

83.2

3–42.7

0

aSo

me

PT

sdesc

ribin

gm

ilddis

ord

ers

were

not

show

n.

bM

ante

l-H

aensz

elad

just

ment

for

co-r

eport

ed

card

iova

scula

rdru

gs(s

ee

supple

menta

rym

ateri

al).

NEC

:N

ot

els

ewhere

codifi

ed.

Roberto et al. 9

In terms of the demographic characteristics of trip-tan reports, the mean age and female/male ratio were inagreement with the epidemiological features ofmigraine (12).

Almost all disproportionality signals obtainedappeared independent from possible underlying cardio-vascular disorders (identified by co-reported cardiovas-cular drugs and used for the statistical adjustment),suggesting that triptans may represent an additionalrisk factor for serious vascular events and confirmingthe need to pay special attention to underlying riskbefore prescribing a triptan. This result is in accordancewith the finding of Visser et al., who found no differ-ence in cardiovascular risk between subjects with andwithout sumatriptan-induced chest symptoms among735 migraine patients who had used sumatriptan fortwo years (22). Therefore, the apparent independenceof our results from the concomitant use of other car-diovascular drugs could substantiate the hypothesis ofa rare subjective susceptibility (e.g. genetic variations)to a triptan vasoconstrictive effect, as also suggested bysome cases of serious ischemic events in patients with-out pre-existing risk factors (6–9).

The main unexpected signals of disproportionatereporting are discussed below.

Cerebrovascular events

We found a significant association between triptan useand a range of serious cerebrovascular events such as‘cerebral infarction’ (patients’ mean age 42.6 years;female/male ratio 3.3:1) and ‘ischaemic stroke’(patients’ mean age 40.1 years; female/male ratio1.8:1). The FDA prescribing information of Imitrex�

(sumatriptan) currently states that ‘cerebrovascularevents have been reported (. . .). The relationship ofsumatriptan to these events is uncertain’ (5). Instead,this information is not mentioned in the relevant sum-mary of product characteristics available in the UnitedKingdom (UK) (10), where the oral formulation of

sumatriptan 50mg can also be purchased without aprescription. Moreover, on the basis of post-marketingepidemiological studies, triptans are not considered toincrease the risk of ischaemic stroke when adequatelyprescribed (2,23). For these reasons, we considered suchassociations unexpected. However, our findings mayhave been influenced by an indication and/or a proto-pathic bias. The association between migraine andstroke has yet to be clarified since (a) migraineurs areconsidered to have about a twofold risk of ischaemicstroke (24), (b) migraine per se is identified as one of thenumerous aetiologies of ‘ischaemic stroke of unusualcauses’ (25) and (c) the symptoms of cerebral ischaemiamay resemble a migraine attack (26). Therefore, in thelight of current knowledge, an accurate diagnosis ofmigraine, excluding all other possible mimicking condi-tions, remains crucial for the safe use of triptans as theirvasoconstrictive effect on cerebral vessels (3,27) may actas a triggering factor for brain infarction (28).Appropriateness of use is one of the crucial aspects ofthe safety profile of triptans. The European MedicineAgency (EMA) recently refused the centralised author-isation for a low-dosage sumatriptan formulation as anon-prescription product (29) because of the ‘theoret-ical’ increased risk of cerebro- and cardiovascularevents related to a misuse of these antimigraine agents.

Aneurysms and artery dissections

The second group of unexpected signals concernsaneurysms and artery dissections, which were high-lighted at any level of specificity of the MedDRA clas-sification tested. In particular, the specific combinationtriptan-carotid artery dissection (patients’ mean age47.6 years; female/male ratio 3:1) was striking. Todate there are no published case reports on this associ-ation. A case report of carotid artery dissection afterergotamine abuse hypothesised a causal relationshipbetween the two (30). A possible role of triptan in trig-gering or precipitating such an event cannot be

Table 3. Reporting odds ratio (ROR) for the Standardised MedDRA Queries (SMQs) considered suitable for the purposes of the

present analysis.

Standardised MedDRA queries (SMQs) Triptan cases ROR CI 95% RORa (adjusted) CI 95%

Myocardial infarction 213 0.91 0.79–1.04 0.94 0.82–1.08

Ischaemic cerebrovascular conditions 208 1.01 0.88–1.15 1.04 0.91–1.20

Ischaemic heart disease 112 1.24 1.03–1.50 1.47 1.21–1.77

Ischaemic colitis 77 0.78 0.63–0.98 0.84 0.67–1.06

Central nervous system haemorrhages and cerebral conditions 11 2.82 1.55–5.11 2.95 1.54–5.47

Cerebrovascular disorders 9 2.33 1.21–4.50 2.25 1.09–4.46

aMantel-Haenszel adjustment for co-reported cardiovascular drugs (see supplementary material).

CI: confidence interval; MedDRA: Medical Dictionary for Regulatory Activities.

10 Cephalalgia 34(1)

excluded in the light of the vasoconstrictive effect oftriptans on the carotid arteries (10,27) and of the multi-factorial aetiology of carotid artery dissection.Nevertheless, an indication and/or protopathic biascould also have influenced our results in this case (31,32).

Pregnancy-related vascular events

Our findings raised potential safety issues concerningmaternal and foetal outcomes related to triptan useduring pregnancy.

Concerning maternal vascular risk, an associationbetween triptan use and ‘pregnancy-associated hyper-tension’, ‘pregnancy-induced hypertension’ and ‘pla-cental infarction’ was found respectively. A possiblepathophysiological role of 5HT1b/d receptors in pre-eclampsia had been already hypothesised by Guptaet al. (33), who also provided evidence of a progressiveincrease in contractile response to sumatriptan duringthe third trimester in isolated umbilical arteries fromnormotensive pregnant women. Therefore, on thebasis of the triptan mechanism of action, such eventscould reasonably be included among the conditions tobe monitored when a triptan is used during pregnancy.In any case, disorders such as pre-eclampsia and gesta-tional hypertension, which are also related to placentalinfarction (34), seem to be more frequent in womenwith migraine (35,36).

The second interesting disproportional associationin pregnancy concerns foetal outcomes: Three reportslisting the event ‘congenital arterial malformations’were found. Although this term refers to very non-specific outcomes in few reports, the issue probablymerits further evaluation. To date, available studieson the safety of triptan use in pregnancy appear to bereassuring, but lack robust evidence (37,38). A preclin-ical study reported an increased incidence of cervi-cothoracic vascular malformation in rabbits exposedto triptans during organogenesis (10), making this asso-ciation potentially plausible.

Strengths and limitations

The first strength of this study is the data source. TheFDA_AERS database has already been widely used forpost-marketing surveillance of drugs and biologics,showing its usefulness for the identification of seriousand rare reactions that resulted in labelling changes andin some cases market withdrawals (14). Second, weapplied the disproportionality approach to differentMedDRA levels and to the appropriate SMQs, toincrease the sensitivity of the analysis for possible detec-tion of unexpected events (20). Third, the statisticaladjustment of the ROR attempted to control the poten-tial confounding factor represented by pre-existing risk,

suggested by co-reported cardiovascular drugs.Although the proxy used could generate misclassifica-tion when some beta-blockers or calcium channelblockers are used for migraine attack prophylaxis, alltriptan reports listing such drugs reported at least oneadditional cardiovascular drug as confirmation ofunderlying disease.

Different factors should also be considered potentialsources of bias for the results obtained.Underreporting, selective reporting and residual dupli-cates represent intrinsic limits of the data analysed (18).Contraindications and precautions concerning cerebro-and cardiovascular events included in triptan packageinserts and labels (5,10) may lead prescribers to over-report such events (notoriety bias). Moreover, sincepeople with migraine are considered to be more atrisk for different ischaemic events and other cardiovas-cular conditions (24,35,36), our analysis could havebeen affected by an indication bias. In addition, aprotopathic bias could have also affected our resultsfor those events easily mistaken for a migraine attack,(e.g. ischaemic cerebrovascular conditions and carotidartery dissection (26,32)).

Conclusions

Although triptans remain a safe antimigraine approachwhen appropriately used, this study highlights anumber of reports of unexpected serious vascularevents that may represent rare adverse reactions to trip-tan intake in ‘real’ clinical practice. Our results show arange of ischaemic cerebrovascular events, aneurysmsand artery dissections, and pregnancy-related vascularevents disproportionally reported after triptan use.Nevertheless, the associations discussed here shouldnot be considered definite evidence of causality.A case-by-case analysis of the individual reports con-taining the unexpected events discussed above is neces-sary to confirm or disprove the plausibility of a causalrelationship between drug and event, and to add usefulevidence for further regulatory decisions on the mostappropriate dispensing regimen of this drug class.Lastly, validated signals from spontaneous reportsshould represent hypotheses to be tested throughlarge-scale analytical studies based on clinical data-bases, in order to obtain information on the estimatedrisk rate of the suspected adverse drug reaction inexposed and unexposed subjects.

Although our findings need to be confirmed, theyraise concerns about triptan use in patients withrisk factors for vascular disorders and during preg-nancy. Furthermore, patients with atypical migrainepossibly related to underlying vascular disorders ormalformations should be investigated in depth beforereceiving a triptan.

Roberto et al. 11

Clinical implications

. Appropriateness of use is one of the crucial aspects of the safety profile of triptans, as in rare instances theirvasoconstrictive effect can cause severe vascular events.

. By analysing the United States Food and Drug Administration FDA Adverse Events Reporting Systemdatabase, this study highlights a number of reports of serious and unexpected vascular events associatedwith triptan use (i.e. ischaemic cerebrovascular events, aneurysms and artery dissections, and pregnancy-related vascular events).

. Although our findings await confirmation, they raise concerns about triptan use in patients with risk factorsfor vascular disorders and during pregnancy. Furthermore, patients with atypical migraine disorders pos-sibly related to underlying vascular conditions or malformation should be investigated in depth beforereceiving a triptan.

Funding

The study was supported by funds from the University ofBologna.

Conflict of interest

None declared.

References

1. Duncan CW, Watson DP and Stein A. Diagnosis and

management of headache in adults: Summary of SIGNguideline. BMJ 2008; 337: a2329.

2. Martin VT and Goldstein JA. Evaluating the safety andtolerability profile of acute treatments for migraine. Am J

Med 2005; 118(Suppl 1): 36S–44S.3. Tfelt-Hansen P, De Vries P and Saxena PR. Triptans in

migraine: A comparative review of pharmacology,

pharmacokinetics and efficacy. Drugs 2000; 60: 1259–1287.4. Dodick D, Lipton RB, Martin V, et al. Consensus state-

ment: Cardiovascular safety profile of triptans (5-HT

agonists) in the acute treatment of migraine. Headache2004; 44: 414–425.

5. GlaxoSmithKline. Imitrex (Sumatriptan succinate) –

FDA prescribing information, http://www.accessdata.fda.gov/spl/data/ba18a310-baaf-4663-a5b7-709b3320d0b2/ba18a310-baaf-4663-a5b7-709b3320d0b2.xml (2011,accessed October 2012).

6. Smith M, Golwala H and Lozano P. Zolmitriptaninduced acute coronary syndrome: A unique case. Am JTher 2011; 18: e153–e156.

7. Hodge JA and Hodge KD. Ischemic colitis related tosumatriptan overuse. J Am Board Fam Med 2010; 23:124–127.

8. Barra S, Lanero S, Madrid A, et al. Sumatriptan therapyfor headache and acute myocardial infarction. ExpertOpin Pharmacother 2010; 11: 2727–2737.

9. Erbilen E, Ozhan H, Akdemir R, et al. A case of myo-cardial infarction with sumatriptan use. Pediatr Cardiol2005; 26: 464–466.

10. ForestLaboratoriesUKLimited. Imigran (Sumatriptan

succinate) – MHRA, Summary of Product Character-istics, http://www.medicines.org.uk/EMC/medicine/751/SPC/ImigranþTabletsþ 50mgþ ImigranþTabletsþ

100mg/ (2012, accessed October 2012).

11. Bigal ME, Golden W, Buse D, et al. Triptan use as a

function of cardiovascular risk. A population-basedstudy. Headache 2010; 50: 256–263.

12. Biagi C, Poluzzi E, Roberto G, et al. Pattern of triptan

use and cardiovascular coprescription: A pharmacoepide-miological study in Italy. Eur J Clin Pharmacol 2011; 67:1283–1289.

13. Panconesi A, Pavone E, Vacca F, et al. Triptans in theItalian population: A drug utilization study and a litera-ture review. J Headache Pain 2008; 9: 71–76.

14. Wysowski DK and Swartz L. Adverse drug event surveil-

lance and drug withdrawals in the United States, 1969–2002: The importance of reporting suspected reactions.Arch Intern Med 2005; 165: 1363–1369.

15. Mozzicato P. MedDRA: An overview of the MedicalDictionary for Regulatory Activities. Pharmaceut Med2009; 23: 65–75.

16. World Health Organisation. WHO Collaborating Centrefor Drug Statistics Methodology – Guidelines for ATCclassification and DDD assignment, 2013. Oslo, 2012.

17. Poluzzi E, Raschi E, Moretti U, et al. Drug-induced tor-sades de pointes: Data mining of the public version of theFDA Adverse Event Reporting System (AERS).Pharmacoepidemiol Drug Saf 2009; 18: 512–518.

18. Poluzzi E, Raschi E, Piccinni C, et al. Data mining tech-niques in pharmacovigilance: Analysis of the publiclyaccessible FDA Adverse Event Reporting System

(AERS), http://cdn.intechopen.com/pdfs/38579/InTech-Data_mining_techniques_in_pharmacovigilance_analysis_of_the_publicly_accessible_fda_adverse_event_reporting_

system_aers_.pdf (2013, accessed October 2012).19. Bate A and Evans SJ. Quantitative signal detection using

spontaneous ADR reporting. Pharmacoepidemiol DrugSaf 2009; 18: 427–436.

20. Pearson RK, Hauben M, Goldsmith DI, et al. Influ-ence of the MedDRA hierarchy on pharmacovigilancedata mining results. Int J Med Inform 2009; 78:

e97–e103.21. Moore N, Hall G, SturkenboomM, et al. Biases affecting

the proportional reporting ratio (PPR) in spontaneous

reports pharmacovigilance databases: The example ofsertindole. Pharmacoepidemiol Drug Saf 2003; 12:271–281.

22. Visser WH, Jaspers NM, de Vriend RH, et al. Chestsymptoms after sumatriptan: A two-year clinical practice

12 Cephalalgia 34(1)

review in 735 consecutive migraine patients. Cephalalgia1996; 16: 554–559.

23. Archambault ME. Do triptans increase the risk of

thromboembolic stroke? JAAPA 2006; 19: 57–58.24. Bigal ME. Migraine and cardiovascular disease. Arq

Neuropsiquiatr 2011; 69: 122–129.25. Arboix A, Massons J, Garcıa-Eroles L, et al. Migrainous

cerebral infarction in the Sagrat Cor Hospital of Barce-lona stroke registry. Cephalalgia 2003; 23: 389–394.

26. Agostoni E, Fumagalli L, Santoro P, et al. Migraine and

stroke. Neurol Sci 2004; 25(Suppl 3): S123–S125.27. Totaro R, De Matteis G, Marini C, et al. Sumatriptan

and cerebral blood flow velocity changes during migraine

attacks. Headache 1997; 37: 635–639.28. Gazioglu S, Boz C and Ozmenoglu M. Basilar artery

occlusion in migraine-like headache: A possible triggering

effect of sumatriptan. Neurol Sci 2012; 33: 125–128.29. European Medicines Agency. Assessment report –

Sumatriptan Galpharm 50mg tablets, http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_

Public_assessment_report/human/002140/WC500122862.pdf (2011, accessed October 2012).

30. Akova-Ozturk E, Husstedt IW, Ringelstein EB, et al.

Carotid artery dissection in ergotamine abuse.Headache 2004; 44: 930–932.

31. Rubinstein SM, Peerdeman SM, van Tulder MW, et al. A

systematic review of the risk factors for cervical arterydissection. Stroke 2005; 36: 1575–1580.

32. Haneline MT and Rosner AL. The etiology of cervicalartery dissection. J Chiropr Med 2007; 6: 110–120.

33. Gupta S, Hanff LM, Visser W, et al. Functional reactiv-

ity of 5-HT receptors in human umbilical cord andmaternal subcutaneous fat arteries after normotensiveor pre-eclamptic pregnancy. J Hypertens 2006; 24:1345–1353.

34. Salgado SS and Pathmeswaran A. Effects of placentalinfarctions on the fetal outcome in pregnancies compli-cated by hypertension. J Coll Physicians Surg Pak 2008;

18: 213–216.35. Bushnell CD, Jamison M and James AH. Migraines

during pregnancy linked to stroke and vascular diseases:

US population based case-control study. BMJ 2009; 338:b664.

36. Facchinetti F, Allais G, Nappi RE, et al. Migraine is

a risk factor for hypertensive disorders in pregnancy:A prospective cohort study. Cephalalgia 2009; 29:286–292.

37. Duong S, Bozzo P, Nordeng H, et al. Safety of triptans

for migraine headaches during pregnancy and breastfeed-ing. Can Fam Physician 2010; 56: 537–539.

38. Nezvalova-Henriksen K, Spigset O and Nordeng H.

Triptan exposure during pregnancy and the risk ofmajor congenital malformations and adverse pregnancyoutcomes: Results from the Norwegian Mother and

Child Cohort Study. Headache 2010; 50: 563–575.

Roberto et al. 13