Background: The relationship between migrainous aura and patent foramen ovale (PFO) remains uncertainMethods: We performed ‘bubble’ contrast transthoracic echocardiography on 80 migraineurs with 4–15headache days per calendar month, mean age 45 (75% female) with mean 9.4 headache days. A large PFOwas defined by passage of an uncountable bolus of bubbles or complete opacification of the left-heart in≤3 cycles, while a moderate PFO was defined by passage of ≥20 bubbles but not sufficient to form a bolusor opacify the whole of the left heart.Results: There was a moderate or large PFO in 28 (35%; 95% CI 24–46%). There were atrial septal aneurysms in11 (15%) and the relationship with PFO was moderately strong (spearman rank correlation 0.493, CI 0.308 to
0.643; p b 0.0001). Seven (9%) had a valve abnormality and one had a closure device for an atrial septaldefect. Aura were reported in 31 (39%) and in these the PFO was large in 9 (29%) compared with 8 (16%;p = 0.143) without aura. There was no relationship between the size of the PFO and the number of headachedays (difference between no PFO and large PFO=−0.6 days; 95% CI –2.6 to +1.4; p-value for any differencebetween groups = 0.316).Conclusions: The prevalence of moderate or large patent foramen ovale was 35% and was not associated withthe presence of aura or the frequency of the headache.
The prevalence and significance of a patent foramen ovale (PFO) inmigraineurs is not known. Some studies suggest a higher prevalencecompared with control subjects [1], but others find no difference [2]. Itis important to establish the prevalence because of a growing trendtowards percutaneous closure. In patients with cryptogenic stroke or de-compression sickness in divers, closure of a PFO [3,4], but not of an ASD[5], has been reported to reduce coexistent migraine. Non-randomisedtrials have suggested that closure is a reasonable treatment for migraineas sole pathology [6–8], but the only randomised trial [9] showed noeffect. This trial was criticised for using insufficiently robust criteria fordiagnosing a PFO [10].
unhill Medical Trust
Thomas' Hospital, London SE128.bers).
e reliability and freedom fromation.
Ltd. All rights reserved.
Association of migraine aura13.06.054
After cryptogenic stroke or TIA, an atrial septal aneurysm associatedwith a PFO increases the risk of future cerebro-vascular events [11]. Thestrength of association between atrial septal aneurysm and migraine,however, remains unknown.
We recruited a series of patients withmigraine referred to headachespecialists for a therapeutic trial. In this paper we describe the preva-lence of PFO, atrial septal aneuryms and other structural abnormalities,and their association with aura and the frequency of headache.
2. Methods
2.1. Patients
Echocardiogramswere performed as part of a randomised trial of clopidogrel vs place-bo as treatment for migraine (ISRCTN 36114412). The results of this trial will be reportedseparately. The inclusion criteria were age N18, definite migraine with or without auraaccording to International Headache Criteria [12], andmigraine (headache or aura) occur-ring on 4 or more days in a 28 day month [13]. Exclusions were high risk featuressuggesting cerebral malignancy or other pathology (e.g. arterio-venous malformation),more than 15headachedays in a 28 day period, contraindications to clopidogrel treatment(hypersensitivity, active bleeding includingmenorrhagia, warfarin therapy, knownhepaticdysfunction), requirement for clopidogrel treatment (e.g. coronary stent) or for routinenon-steroidal anti-inflammatory agent or aspirin other than for acute headache, use ofan investigational product within 3 months, inability to understand English, pregnancyor breast-feeding, clinically significant abnormalities of platelet or liver function.
with patent foramen ovale and atrial septal aneurysms, Int J Cardiol
2 J. Chambers et al. / International Journal of Cardiology xxx (2013) xxx–xxx
2.2. Headache
The frequency of headache was calculated as headache days using a diary filled outfor 28 days before randomisation. For a headache or aura lasting b24 h, one headacheday was counted. For a headache or aura lasting 24–48 h, two headache days werecounted. The patient was asked to indicate which headaches were associated with anaura. The patient was taken as having migraine with aura if one or more headachewas associated with aura. If no headaches were associated with aura, the patient wasdefined as having migraine without aura.
2.3. Echocardiography
Participants had echocardiography after randomisation, but at no fixed visit. A fullstandard transthoracic study was performed. Zoomed views of the atrial septum wereobtained in the apical 4-chamber and subcostal views and an off-axis 4-chamber viewwith the transducer moved medially within the 5th space to better show the septum. Asearch for a left-to-right atrial shunt on colour mapping was made. The size of the rightand left atria were measured. A bubble study was then performed, by one operator (JC)in all but one case which was instead performed by a sonographer trained to conduct aspecialist sonographer-led clinical bubble list. A 21G cannula was positioned in anantecubital fossa vein usually on the left. This was then connected to a three-way
Fig. 1. Examples of patent foramen ovale. Examples of a large shunt with
Please cite this article as: Chambers J, et al, Association of migraine aura(2013), http://dx.doi.org/10.1016/j.ijcard.2013.06.054
tap. One dry syringe and 6 syringes filled with 7–8 ml N saline were drawn up. Foreach injection, approximately 0.5 ml air was left in the syringe and 0.5 ml venousblood was drawn back into the syringe. The dry syringe was then attached to theother port of the 3-way tap and the mixture agitated between the two syringes untila dense froth containing no air bubbles was obtained. This was injected with nodelay. For the first injection the subject was asked to produce a short, sharp cough un-less there was already spontaneous passage of microcavitations (‘bubbles’). All otherinjections were made with a Valsalva manoeuvre. The patient was instructed to breathout then hold the breath and strain against a closed glottis with minimal movement ofthe chest. This was practised until the effort and release were optimal with deviation ofthe septum to the left but with minimal disturbance to the image quality. Up to 5 injec-tions with Valsalva were given unless dense opacification of the left heart occurredsooner [11].
2.4. Echocardiographic analysis
A PFO was said to be present for this study if there was passage of bubbles into theleft atrium within 3 or fewer cycles after the injection. The PFO was classed [1] as: largeif there was light or dense opacification of the left heart (Fig. 1A, B) or passage of a largebolus of bubbles too dense for counting; moderate if there were ≥20 bubbles, but nobolus or opacification; and small if there were b20 bubbles.
light (A) and dense (B) opacification of the whole of the left heart.
with patent foramen ovale and atrial septal aneurysms, Int J Cardiol
3J. Chambers et al. / International Journal of Cardiology xxx (2013) xxx–xxx
For comparison with the literature, an analysis was also performed using the samebubble criteria numbers but a cut-off of ≤5 cycles rather than ≤3 cycles [14,15]. Wealso used the criteria as used in MIST [9] with a small shunt classed as 1–5 bubbles, amoderate shunt as 6–20 bubbles and a large shunt N20 bubbles seen within the leftatrium in ≤5 cycles. The rationale for the different grading criteria adopted for thisstudy is that, probably as a result of advances in machine technology it is common tosee dense opacification of the whole left heart and it is likely that this denotes ashunt which is larger than one causing 20 [14] or 25 [15], 30 [2] or even 50 bubbleswhich are previously-described cut-points for a significant shunt.
An atrial septal aneurysmwas defined by a base ≥10 mmwide and an apex movingduring spontaneous respiration between left and right atria with an excursion ≥10 mm(Fig. 2) [11,16,17]. A mobile septum was defined by an excursion of b10 mm. A normalseptum was defined by no movement or trivial movement. All abnormalities werenoted including the presence of a bicuspid aortic valve or mitral prolapse, aortic valvethickening and mitral annular calcification.
2.5. Statistical analysis
Mean values were calculated for age and headache days per calendar month. Therelationship between PFO and atrial septal mobility was tested using a Spearman'srank test. The relationship between PFO and aura or headache days was tested withthe Wilcoxon rank-sum test.
Fig. 2. Atrial septal aneurysm. The atrial septum in its most r
Please cite this article as: Chambers J, et al, Association of migraine aura(2013), http://dx.doi.org/10.1016/j.ijcard.2013.06.054
3. Results
A total of 80 participants had echocardiography between 14 May2009 and 28 June 2012. The mean age was 45 (range 18–68) and 62(75%) were female. The mean number of headache days was 9.4 (range4–14).
3.1. Incidence of echocardiographic abnormalities
Echocardiograms with bubble studies were successfully performedin all 80 subjects. There was a moderate or large PFO in 28 (35%; 95%CI 24–46%) (Table 1).
Using the broader criterion of bubbles appearing within ≤5 cycles,the number of moderate or large PFOs increased by only one althoughthe number of small PFOs increased from 5 (6%) to 11 (15%). Usingthe MIST criteria, a larger proportion, 43%, had moderate or large PFOs.
Therewere atrial septal aneurysms in 11 (15%)whichwere associat-ed with a large PFO in 6 (55%) (Table 2). There was a mobile septum in
ightward (A) and its most leftward (B) positions at rest.
with patent foramen ovale and atrial septal aneurysms, Int J Cardiol
⁎ Ashunt was large if there was light or dense opacification of the left heart (Fig. 1A, B)or passage of a large bolus of bubbles too dense for counting; moderate if there were≥20bubbles, but no bolus or opacification; small shunt if there were 1–20 bubbles.⁎⁎ MIST criteria [9]: Small shunt 1–5 bubbles, moderate shunt 6–20 bubbles, large shuntN20 bubbles within the left atrium within 5 or fewer cycles.
Table 3Association of aura and presence or size of pfo.
4 J. Chambers et al. / International Journal of Cardiology xxx (2013) xxx–xxx
11 (15%) which was associated with a large PFO in 3 (27%) (Table 2).The relationship between PFO and mobility was moderately strong(Spearman's rank correlation = 0.493, CI 0.308 to 0.643; p b 0.0001).
Seven (9%) had a valve abnormality, a bicuspid aortic valve in 3(4%), mitral prolapse in 2 (2.5%), mitral annular calcification in one(1.3%) and mild aortic valve thickening in one (1.3%). These were as-sociated with no PFO in 3, a moderate PFO in one and large PFO in 3. Afurther case had a closure device for an atrial septal defect and thiswas associated with a moderate passage of bubbles.
3.2. Association between aura and frequency of headaches
Aura were reported in 31 (39%) cases. There was no evidence oflarger PFOs in patients with aura (Table 3; p = 0.143 by Wilcoxonrank-sum test). There was likewise no evidence of differences in theproportion of people with mobile or aneurysmal atrial septa andthose with aura compared with no aura (Table 4).
There was no relationship between the size of the PFO and thenumber of headache days. The difference between no PFO and largePFO was −0.6 days (95% CI −2.6 to +1.4; p-value for any differencebetween groups = 0.316).
4. Discussion
This study showed a 35% prevalence of moderate or large patentforamina ovale (PFO). A PFO was not more common with aura thanwithout aura, and was not related to the frequency of headache. Anatrial septal aneurysm was found in 15% of patients, and was notmore frequent with aura. There was a valve abnormality present in7 (9%), most commonly a bicuspid aortic valve.
The relationship between PFO andmigraine has been difficult to de-termine. Early studies included migraine associated with cerebral in-farction or decompression sickness as the main reason for recruitment[4]. This may have introduced bias in the study population. In a system-atic review [3], the prevalence of PFO was 41–72% in patients withmigraine and aura, but 16–34% without aura. This range contains theprevalence shown in the present study. However, in a case-controlstudy excluding patients with prior stroke [2] the incidence of PFOwas similar in patients with migraine with or without aura and also incontrols. In a large population-based study [18] of 1101 patients whohad not had a stroke, the incidence of PFO was 15% for those both
Table 2Association between IAS mobility and pfo.
PFO size Normal IAS N = 58 Mobile IAS N = 11 Aneurysmal IAS N = 11
Please cite this article as: Chambers J, et al, Association of migraine aura(2013), http://dx.doi.org/10.1016/j.ijcard.2013.06.054
with and without migraine. The incidence was slightly higher, 16% inthose with aura compared with 11% in those without aura.
These differences may have arisen from variations in selectioncriteria or in measurement methods. The cut-point for a significantPFO is variously taken as 10 bubbles, [19], 20 bubbles [15,20–22], 25or 30 bubbles [2,10] or occasionally 50 bubbles [23]. However theseare relatively similar numbers particularly bearing inmind the difficultyof counting individual bubbles. It is common for too many bubbles tocount to cross as a bolus or fill the whole of the left atrium and left ven-tricle either lightly (Fig. 1A) or densely (Fig. 1B) and these are graded aslarge. In this we agree with others [1]. We grade a shunt as moderate if≥20 bubbles cross but not enough to opacify the whole left heart. Amore important difference is in the timing. We allowed bubbles onlywithin 3 or fewer cycles [2,24] while other studies allow 4 or fewer[15] or 5 or fewer cycles [9]. Allowing 4–5 cycles is likely to increasethe false positive rate by including transpulmonary shunting. Inthe MIST trial, which used ≤5 cycles, a PFO could not be found atattempted closure in 5 of 74 patients. Reanalysing our results usingthe MIST criteria we obtained an incidence of moderate or large PFOof 43% rather than 36%. There are other differences. Studies vary in thenumber of injections. We used up to 6 although large shunts were usu-ally obvious with fewer. We used N-saline mixed with air and bloodwhile some use gelofuscin and this may produce more bubbles. Weused transthoracic rather than transoesophageal echocardiography. Anumber of comparisons show that sensitivities are similar for the twotechniques [24–27], and sometimes slightly better for transthoracicechocardiography [15].
There is little information on atrial septal aneurysms. The septummoves more with coughing or a Valsalva manoeuvre than with quietrespiration and there is a continuous spectrum in excursion at rest orwith manoeuvres. The difference between a mobile and an aneurymalseptum is defined by an arbitrary, but frequently used [17], cut-off ex-cursion of≥10 mm.We found a mobile septum in 11 (15%) and an an-eurysmal septum also in 11 (15%). Garg et al. [2] showed a similarincidence of hypermobile septa in 9% of migraineurs compared with7.6% of controls. However the same study found no aneurysmal septa.Other series also showed a far lower incidence, 1.7% [28] or 2.5% [29]than in our study. The atrial septum is in the far-field of a transthoracicechocardiogram and its accurate delineation depends on good imagequality which is more likely with modern instruments. We showedthat 75% of atrial septal aneurysms were associated with a moderateor large PFOs which is similar to previous work in patients after crypto-genic stroke [29,30].
It is possible that platelet aggregates can form on a mobile or aneu-rysmal atrial septum and either vasoactive substances formed or theplatelets themselves may then pass across a PFO. In patients after cryp-togenic stroke, one study found that the incidence of a further eventwas significantly higher if there was a combination of an atrial septalaneurysm and PFO rather than aneurysm or PFO alone [11]. This wasnot confirmed in another study [16] andwhether a similar effect occursin migraine remains speculative.
Table 4Association between aura and septal mobility.
5J. Chambers et al. / International Journal of Cardiology xxx (2013) xxx–xxx
The higher than expected incidence of heart valve abnormalitiesparticularly bicuspid aortic valve has not previously been shown in mi-graine. The incidence of bicuspid aortic valvewas 3.7% comparedwith apopulation prevalence of 0.5–0.8% [31,32]. The incidence of mitral pro-lapse was 2.5% which is similar to the expected population prevalenceof 2% [33,34]. Bicuspid valves andmitral prolapse can rarely be associat-ed with platelet aggregation causing stroke or peripheral emboli. It ispossible that they are also related to migraine via platelet function asfor atrial septal aneurysms, either as a result of emboli or release of va-soactive substances from aggregating platelets.
4.1. Limitations
Our conclusions are limited by the sample size and the fact that wehad no control groupwithout headache.We note that, at 35%, our prev-alence is higher than found in population-based studies, using differentmethodologies: 15% using echocardiography [18], 27% at autopsy [35]and 24% at cardiac surgery [28]. However the different methods usedlimit the values of such comparisons.
5. Conclusion
This study shows a high incidence of structural abnormalities onechocardiography in subjects with 4–15 headache days per calendarmonth. The prevalence of patent foramen ovale was affected by thereporting criteria. A patent foramen ovale was not statistically morecommon with aura and was not associated with the frequency of theheadache.
Acknowledgement
We thank the nurses who collected the data for his study, Sr ElaineWong, Thanuja Weerasinghe, Azanu Golda-Grace.
We thank the following for helpingwith recruitment: Dr Sam Chong,Ranjan Das, Dr Jane Docherty, Dr Tyrrel Evans, Dr Raj Mitra, Dr RobWeeks, Dr Rachael Kilner, Dr Alex Pothen and the staff at Albion StreetSurgery, Crofton Surgery, Bromley Common Surgery, Streatham Com-mon Surgery and the Hurley Clinic,
We also thankmembers of the steering committee: Prof Anish Bahra,Dr Richard Peatfield and the Data Monitoring Committee: Dr RichardPeatfield and Prof Nicola Crichton
Sanofi donated the clopidogrel and placebo for the therapeutic trialbut took no part in the design or analysis of the main, or any substudy
References
[1] SchwerzmannM,NedeltchevK, Lagger F, et al. Prevalence and size of directly detect-ed patent foramen ovale in migraine with aura. Neurology 2005;65:1415–8.
[2] Garg P, Servoss SJ, Wu JC, et al. Lack of association between migraine headacheand patent foramen ovale. Circulation 2010;121:1404–12.
[4] Wilmshurst PT, Nightingale S, Walsh KP, et al. Effect on migraine of closure of car-diac right-to-left shunts to prevent recurrence of decompression illness or strokeor for haemodynamic reasons. Lancet 2000;356:1648–51.
[5] Mortelmans K, Post M, Thijs V, Herroelen L, Budts W. The influence of percutaneousatrial septal defect closure on the occurrence of migraine. Eur Heart J 2005;26:1533–7.
[6] SchwerzmannM,Wiher S, NedeltchevK, et al. Percutaneous closure of patent foramenovale reduces the frequency of migraine attacks. J Am Coll Cardiol 2006;47:446–8.
[7] Azarbal B, Tobis J, Suh W, Chan V, Dao C, Gaster R. Association of interatrial shuntsand migraine headaches. JACC 2005;45:489–92.
Please cite this article as: Chambers J, et al, Association of migraine aura(2013), http://dx.doi.org/10.1016/j.ijcard.2013.06.054
[8] Reisman M, Christofferson RD, Jesurum J, et al. Migraine headache relief aftertranscatheter closure of patent foramen ovale. JACC 2005;45:493–5.
[9] Dowson A, Mullen MJ, Peatfield R, et al. Migraine Intervention with STARFlex Tech-nology (MIST) Trial. A prospective, multicenter, double-blind, sham-controlled trialto evaluate the effectiveness of patent formaen ovale closure with STARFlex septalrepair implant to resolve refractory migraine headache. Circulation 2008;117:1397–404 [Correction Circulation 2009;120:e71-2].
[10] Johansson MC, Dellborg M, Eriksson P. Letter on ‘migraine intervention withSTARFlex Technology (MIST) trial’. Circulation 2009;1109:e193.
[11] Mas JL, Arquizan C, Lamy C, et al. Recurrent cerebrovascular events associated withpatent foramen ovale, atrial septal aneurysm, or both. N Engl J Med 2001;345:1740–6.
[12] Headache Classification Committee of the International Headache Society. Classifica-tion and diagnostic criteria for headache disorders, cranial neuralgias and facial pain.Cephalgia 1988;8(Suppl. 7):1–96.
[13] Ridsdale L, Clark LV, Dowson AJ, et al. How do patients referred to neurologists forheadache differ from those managed in primary care? Brit J Gen Prac 2007;57:1–8.
[14] Webster MW, Chancellor AM, Smith HJ, et al. Patent foramen ovale in youngstroke patients. Lancet 1988;2:11–2.
[15] Daniels Z, Weytjens C, Cosyns B, et al. Second harmonic transthoracic echocardio-graphyL the new reference screening method for the detection of patent foramenovale. Europ J Echo 2004;5:449–52.
[16] Homma S, Di Tullio MR, Sacco RL, Mihalatos D, Mandri JL, Mohr JP. Characteristics ofpatent foramen ovale associated with cryptogenic stroke: a biplane transesophagealechocardiographic study. Stroke 1994;25:582–6.
[17] Mugge A, Daniel WG, Kloepper JW, Lichtlein PR. Visualization of patent foramenovale by transesophageal color-coded Doppler echocardiography. Am J Cardiol1988;62:837–8.
[18] Rundek T, Elkind MSV, Di Tullio MR, et al. Patent foramen ovale and migraine. Across-sectional study from the Northern Manhattan Study (NOMAS). Circulation2008;118:1419–24.
[19] Cabanes L, Coste J, Derumeaux G, et al. Interobserver and intraobserver variability indetection of patent foramen ovale and atrial septal aneurysm with transesophagealechocardiography. J Am Soc Echocardiogr 2002;15:441–6.
[20] Braun M, Fassbender D, Schoen SP, et al. Transcatheter closure of patent foramenovale in patients with cerebral ischaemia. JACC 2002;39:2019–25.
[21] Wahl A, Praz F, Tai T, et al. Improvement of migraine headaches after percutaneousclosure of patent foramen ovale for secondary prevention of paradoxical embolism.Heart 2012;96:967–73.
[22] Wilmshurst PT, Nightingale S,Walsh KP, MorrisonWL. Clopidogrel reduces migrainewith aura after transcatheter closure of persistent foramen ovale and atrial septal de-fects. Heart 2005;91:1173–5.
[23] Sztajzel R, Genou D, Roth S, Mermillod B, Le Flochrohr J. Patent foramen ovale, apossible cause of symptomatic migraine: a study of 74 patients with acuteischaemic stroke. Cerebrovasc Dis 2002;13:102–6.
[24] Kuhl HP, Hoffmann R, Merx MW, et al. Transthoracic echocardiography using sec-ond harmonic imaging. J Am Coll Cardiol 1999;34:1823–30.
[25] Souteyrand G, Motreff P, Lusson J-R, et al. Comparison of transthoracic echocardiog-raphy using second harmonic imaging, transcranial Doppler and transesophagealechocardiography for the detection of patent foramen ovale in stroke patients.Europ J Echo 2006;7(2):147–54.
[26] SloanMA, Alexandrov AV, Tegeler CH. Therapeutics and Technology Assessment Sub-committee of the American Academy of Neurology. Assessment: Transcranial Dopplerultrasonography: report of the Therapeutics and Technology Assessment Subcommit-tee of the American Academy of Neurology. Neurology 2004;62:1468–81.
[27] Clarke NR, Timperley J, Kelion AD, Banning AP. Transthoracic echocardiographyusing second harmonic imaging waith valsalva manoeuvre for the detection ofright to left shunts. Eur J Echo 2004;5:176–81.
[28] Krasuski RA, Hart SA, Allen D, et al. Prevalence and repair of intraoperatively diag-nosed patent foramen ovale and association with perioperative outcomes andlong-term survival. JAMA 2009;302(3):290–7.
[29] Di Tullio MR, Sacco RL, Sciacca RR, Jin Z, Homma S. Patent foramen ovale and therisk of ischaemic stroke in a multiethnic population. JACC 2007;49:797–802.
[30] Cabanes L, Mas JL, Cohen A, et al. Atrial septal aneurysm and patent foramen ovaleas risk factors for cryptogenic stroke in patients less than 55 years of age. Stroke1993;24:1865–73.
[31] Nistri S, Basso C, Marzari C, et al. Frequency of bicuspid aortic valve in conscriptsby echocardiogram. Am J Cardiol 2005;96:718–21.
[32] Movahed MR, Hepner AD, Ahmadi-Kashani M. Echocardiographic prevalence ofbicuspid aortic valve in the population. Heart Lung Circ 2006;15:297–9.
[33] Freed LA, Levy D, Levine RA, et al. Prevalence and clinical outcome of mitral-valveprolapse. New Engl J Med 1999;341:1–7.
[34] Sutton M, St J, Weyman AE. Mitral valve prolapse prevalence and complications.Circulation 2002;106:1305–7.
[35] Hagen PT, Scholz DG, Edwards WD. Incidence and size of patent foramen ovaleduring the first 10 decades of life: an autopsy study of 965 normal hearts. MayoClin Proc 1984;59:17–20.
with patent foramen ovale and atrial septal aneurysms, Int J Cardiol
