Conflict of Interest: None
Incomplete Posterior Circle of Willis: A Risk Factor for Migraine?
Article first published online: 26 MAR 2009
© 2009 the Authors. Journal compilation © 2009 American Headache Society
Headache: The Journal of Head and Face Pain
Volume 49, Issue 6, pages 879–886, June 2009
How to Cite
Bugnicourt, J.-M., Garcia, P.-Y., Peltier, J., Bonnaire, B., Picard, C. and Godefroy, O. (2009), Incomplete Posterior Circle of Willis: A Risk Factor for Migraine?. Headache: The Journal of Head and Face Pain, 49: 879–886. doi: 10.1111/j.1526-4610.2009.01389.x
- Issue published online: 27 MAY 2009
- Article first published online: 26 MAR 2009
- Accepted for publication December 20, 2008.
- vascular disorders;
- circle of Willis
Background.— Migraine is associated with vascular risk factors and white matter abnormalities (WMA). Cerebral hypoperfusion is known to be one mechanism underlying WMA and a few studies have shown that an incomplete circle of Willis (CW) may predispose to cerebral hypoperfusion. This study assessed the relationship between the morphologic characteristics of the CW and migraine.
Methods.— This case–control study was carried out in the Amiens University Hospital. Patients undergoing 3-dimensional time of flight magnetic resonance angiography of the CW from January 1 to June 30, 2006 were included (n = 124). A definitive diagnosis of migraine was established in 47 patients: 23 (48.9%) experienced migraine without aura and 24 (51.1%) migraine with aura. The remaining 77 patients with other neurologic disorders constituted the control group. The posterior CW was graded as complete when both posterior communicating arteries and the P1 segments of the posterior cerebral artery were present on visual examination and incomplete when one of these vessels was missing (interobserver agreement: Ktotal = 0.746).
Results.— Incomplete posterior CW was significantly more common in migraineurs than in the control group (49% vs 18%; P < .001). On multivariate analysis, incomplete posterior CW was the sole independent factor associated with migraine (OR: 6.5; 95% CI: 2.6-16.2; P < .001). No difference was found between migraineurs with and without aura.
Conclusions.— Despite some methodological limitations, our results showed that incomplete posterior CW was associated with migraine.
The circle of Willis (CW) is the main cerebral collateral pathway that allows flow redistribution in response to low perfusion pressure in a major feeding artery.1 The posterior communicating artery is the main pathway from the anterior to the posterior circulation (Fig. 1). Angiographic studies have shown that the CW is complete in only 20% of subjects,2 and a few studies have shown that an abnormal CW (a circle with hypoplastic or missing segments) affects regulation of regional cerebral blood flow,3,4 especially in the posterior territory.5
Several studies have shown a strong relationship between migraine and cerebral ischemia. Patients with migraine have an increased prevalence of subclinical cerebral infarcts and silent white matter abnormalities6 attributed to cerebral hypoperfusion accompanying the cortical spreading depression (CSD). Migraineurs have also an increased prevalence of patent foramen ovale,7 coagulation activation,8 and vascular dysregulation.9 Furthermore, migraine-like headache is observed in cerebral angiopathies such as cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy,10 cervical artery dissection,11 and moyamoya disease.12 This suggests that cerebral angiopathy and ischemia may trigger the pain mechanisms underlying headache and CSD. The relationship between migraine and cerebral ischemia is complex and bidirectional.13
The presence of CW anomalies may contribute to the impaired regulation of cerebral blood flow in response to physiological activation known to occur in migraineurs.14 These data converge to suggest a possible relationship between CW anomalies and migraine. Interestingly, both are prominent in the posterior cerebral circulation: CW anomalies are more frequent in the posterior circulation15 and cerebral infarcts of migraineurs preferentially affect the posterior territory.6
However, the hypothesis that CW may be a risk factor for migraine has rarely been addressed. There is a reported association between CW anomalies and complicated migraine.16 One study found a nonsignificant absolute excess of 5% in the frequency of fetal-type posterior cerebral artery (PCA) (defined as absence of the P1 segment of the PCA, which implies that the PCA is supplied by the ipsilateral internal carotid artery [ICA] via the posterior communicating artery, Fig. 1) in patients with migraine with aura compared with matched controls.17 Additionally, Lovrencic-Huzjan et al found a significantly greater prevalence of vertebral artery hypoplasia in patients with migraine with aura compared with migraine without aura and controls.18
We examined the hypothesis that CW anomalies contribute to migraine susceptibility. The aim of this case–control study was to investigate the relationship between CW morphologic characteristics and migraine.
Subjects.— The study was performed in all patients undergoing brain magnetic resonance imaging (MRI) with MR angiography (MRA) in the radiology department of Amiens University Hospital from January 1 to June 30, 2006. Patients <18 or >80 years old were excluded, as were those with aneurysm, carotid artery occlusion or stenosis >70%, acute stroke, and headache observed during or after a cerebrovascular event, as these conditions might alter the configuration of the CW.
A total of 564 consecutive patients were assessed by MRI and MRA during the study period and 440 were excluded (Fig. 2). In the remaining 124 patients, the following variables were prospectively collected according to a standardized questionnaire for hospitalized patients or by means of a structured phone interview by a trained neurologist for outpatients: age, gender, main comorbidities, vascular risk factors (previously identified or diagnosed during hospitalization) including hypertension (antihypertensive treatment or systolic blood pressure >140 mm Hg or diastolic blood pressure >90 mm Hg), diabetes mellitus (insulin or oral hypoglycemic treatment or fasting blood glucose >7 mmol/L on 2 occasions), hypercholesterolemia (lipid-lowering treatment or total cholesterol >5.2 mmol/L), current smoking, current alcohol consumption (>2 drinks daily), hormone therapy, oral contraceptives, and migraine. The diagnosis of migraine without aura and migraine with aura was based on the second edition of the International Classification of Headache Disorder (ICHD-2) criteria.19 The classification of migraine with aura subtypes was also based on the ICHD-2. In patients with multiple auras subtypes, the most frequent was recorded. Informed consent was obtained from all patients and the study was approved by the local ethics committee (Comission d'Evaluation Ethique des Recherches Non Interventionnelles).
Cerebral MRI.— Cerebral imaging was performed on a 1.5 T MRI machine and always included axial T1, T2, and FLAIR-weighted sequences. CW MRA was performed using a 3-dimensional (3D) time-of-flight (TOF) sequence. The original MRA sequence for each patient was retrieved from an electronic database (DXMM®). Two investigators blinded to clinical data rated the presence of CW segments on MRA.
Definitions of CW Morphology.— All CW vessels visualized on individual axial 3D TOF slices as continuous hyperintense segments were considered to be present. Vessel diameter was not taken into account, as the lower limit of vessel diameter remains arbitrary and affects the prevalence of complete CW.
The anterior CW was graded as complete when both A1 segments of the anterior cerebral arteries and the anterior communicating artery were present (Fig. 1). The posterior CW was graded as complete when both posterior communicating arteries and the P1 segments of the PCA were present (Fig. 1). Conversely, when one of these vessels was missing, the posterior CW was graded as incomplete. The CW was graded as fetal-type PCA when the PCA was supplied by the ipsilateral ICA via the posterior communicating artery, instead of the basilar artery (Fig. 3). In this case, CW was graded incomplete (Fig. 3). Special care was taken to distinguish the posterior communicating arteries from the anterior choroidal arteries by scrolling through the sections and assessing the courses of the arteries in sequential display.
Statistical Analysis.— Kappa values20 were calculated to determine interrater agreement for analysis of CW morphology. Variables associated with migraine were examined in univariate (Student t-tests for continuous variables and chi-square tests with Yates correction for categorical variables) and multivariate analysis using a stepwise regression analysis. The variables subjected to analysis were age, gender, arterial hypertension, diabetes mellitus, hypercholesterolemia, smoking, alcohol, oral contraceptives, and CW morphology (incomplete CW). Only variables reaching a P < .25 in univariate analysis were entered in the logistic regression. P values < .05 were considered to be significant. A second series of analysis compared variables in migraineurs according to the presence of aura. Statistical analyses were performed with SPSS statistical software.
The indications for brain MRI are reported in Table 1. Definite migraine was diagnosed in 47 subjects: 23 (48.9%) without aura and 24 (51.1%) with aura. The remaining 77 patients were treated as controls. The study population was middle-aged with a sex ratio favoring females (Table 2).
|Migraineurs n (%)||Nonmigraineurs n (%)|
|Atypical migraine episode:||30 (63.9)||–|
|Prolonged aura >1 hour||21||–|
|Brain injury||2 (4.3)||1 (1.3)|
|Brain tumor||1 (2.1)||11 (14.3)|
|Psychiatric episode||1 (2.1)||15 (19.5)|
|Multiple sclerosis||–||3 (3.9)|
|Cluster headache||–||8 (10.4)|
|Intracranial infection||–||3 (3.9)|
|Cerebral venous thrombosis||–||5 (6.5)|
|Systemic disease||–||14 (18.1)|
|Seizure||1 (2.1)||9 (11.7)|
|All subjects (n = 124) %||Controls (n = 77) %||Migraineurs (n = 47) %||P value|
|Age (years)||38.7 ± 14.8||40.4 ± 15.4||35.7 ± 13.4||.07|
|Number of males||38 (30.6)||27 (35.1)||11 (23.4)||.23|
|Arterial hypertension||19 (15.3)||12 (15.6)||7 (14.9)||1.00|
|Diabetes mellitus||4 (3.2)||4 (5.2)||0 (0)||.30|
|Dyslipidemia||9 (7.3)||4 (5.2)||5 (10.6)||.30|
|Current smoking||35 (28.2)||19 (24.7)||16 (34.0)||.30|
|Oral contraceptives||28 (22.6)||12 (15.6)||16 (34.0)||.03|
|Circle of Willis (CW)|
|Anterior incomplete||7 (5.6)||4 (5.2)||3 (6.4)||1.00|
|Posterior incomplete||37 (29.8)||14 (18.2)||23 (48.9)||<.001|
|Incomplete CW||41 (33.1)||17 (22.1)||24 (51.1)||.001|
|Fetal configuration||14 (11.3)||11 (14.3)||3 (6.4)||.26|
The interrater agreement for analysis of the posterior CW was good to excellent: the ktotal value was 0.746 (right posterior communicating artery: k = 0.7; left posterior communicating artery: k = 0.97).
An incomplete CW was observed in 37 (29.8%) patients and mainly involved the posterior part. Univariate analysis indicated that incomplete posterior CW was significantly more frequent in migraineurs than in controls (Table 2). The frequency of fetal-type PCA did not significantly differ between migraineurs and control subjects. Multivariate analysis identified only the presence of incomplete posterior CW (OR: 6.5; 95% CI: 2.6-16.2; P < .001) as a factor associated with migraine (unselected factors: age [OR: 0.97; 95% CI: 0.94-1.01; P = .09], gender [OR: 0.76; 95% CI: 0.29-2.04; P = .6], oral contraceptives [OR: 2.60; 95% CI: 0.89-7.61; P = .08]). In order to control for the possible confounding effect of age and sex, multivariate analysis was repeated first in the subgroup of men (n = 38) and second in the subgroup of older patients (n = 61), defined by a median age of 37 years. Both analyses provided the same results: the presence of CW anomalies was an independent factor associated with migraine (male subgroup – OR: 5.3; 95% CI: 1.14-24.47; P = .033; patients over 37 years old – OR: 5.8; 95% CI: 1.80-18.42; P = .003).
There was no relationship between CW abnormalities and the presence or absence of an aura (Table 3). Although incomplete posterior CW was more frequent in aura, the difference did not reach significance.
|All migraineurs (n = 47) (%)||Without aura (n = 23) (%)||With aura (n = 24) (%)||P value|
|Age (years)||35.7 ± 13.4||37.7 ± 16.1||33.8 ± 10.2||.33|
|Gender (no. of men, %)||11 (23.4)||5 (21.7)||6 (25)||1.00|
|Hypertension||7 (14.9)||4 (17.4)||3 (12.5)||.70|
|Dyslipidemia||5 (10.6)||2 (8.7)||3 (12.5)||1.00|
|Smoking||16 (34.0)||5 (21.7)||11 (45.8)||.1|
|Oral contraceptives||16 (34.0)||7 (30.4)||9 (37.5)||.76|
|Circle of Willis (CW)|
|Anterior incomplete||3 (6.4)||1 (4.2)||2 (8.7)||1.00|
|Posterior incomplete||23 (48.9)||9 (37.5)||14 (60.9)||.25|
|Incomplete CW||24 (51.1)||9 (37.5)||15 (65.2)||.15|
In this case–control study, the presence of an incomplete posterior CW was an independent factor for migraine both with and without aura.
Contrary to previous reports, we found a high proportion of complete CW. Marked variations in the frequency of incomplete posterior CW have been reported in previous studies, and several factors may account for this discrepancy including differences in study populations, methods of CW assessment, and criteria for incomplete CW. First, the study populations to date have been heterogeneous and usually included subjects with cerebrovascular diseases,21-23 who are more prone to incomplete CW. Both age and sex influence the frequency of complete CW: in MRA studies, the frequency of complete CW decreases with age (because the lower cerebral blood flow velocity in the elderly results in signal void24) and increases with female sex.15 Our population was characterized by a mean age around 40 years and a female predominance, and this may account for the low frequency of incomplete CW. We therefore performed additional analyses in the subgroups of older patients and men and found that the link between CW anomalies and migraine persisted, which strengthens the relationship between CW anomalies and migraine. Second, methods of CW assessment differ, as the oldest studies were based on autopsy examination or conventional angiography but MRA, a safe technique with reproducible results,25 has been used in more recent studies.5,25-27 Third, criteria for incomplete CW differ: some studies, including ours, diagnosed incomplete CW only in subjects without visible arteries, whereas other studies used vessel diameter as the criterion, which increases the percentage of incomplete CW because a visible circulatory artery of small diameter is considered to be absent.15
There are some limitations of our study. Patients were not randomly selected from the general population but were recruited on the basis of complaints such as severe headache, causing them to attend the emergency department. This and the exclusion of stroke patients account for the large proportion of migraineurs in the population studied, and particularly migraine with aura. Another limitation concerns the evaluation of small intracranial arteries with 3D TOF MRA. Although a good agreement between MRA and conventional angiography has been reported,24 vessels with a slow flow may not show high signal intensity because of a saturation effect;15 consequently, an artery may be erroneously classified as absent. Nevertheless, the high rate of complete CW in our study minimizes this technical bias.
Three mechanisms may, theoretically, account for the relationship between incomplete CW and migraine. First, incomplete CW may impair the regulation of cerebral perfusion, which may trigger headache and CSD. Second, incomplete CW may facilitate direct activation of pain mechanisms by arteries abnormalities. Finally, incomplete CW could reflect a generalized increase in arterial tone induced by migraine. Tietjen,28 from an original work by Vanmolkot et al,29 recently argued for alterations in both systemic and cerebral vasoreactivity in migraineurs. This altered vasoreactivity would decrease cerebral flow arteries, and an artery could therefore be erroneously classified as being absent.
In conclusion, despite some methodological limitations, we found an association between the presence of incomplete posterior CW and a diagnosis of migraine, suggesting that this abnormality may contribute to migraine development. These preliminary results require further studies in a large sample from the general population to determine the exact role of CW abnormalities.
- 5Relationship between variations in the circle of Willis and flow rates in internal carotid and basilar arteries determined by means of magnetic resonance imaging with semiautomated lumen segmentation: Reference data from 125 healthy volunteers. Am J Neuroradiol. 2006;27:1770-1775., , , et al.
- 9Association between migraine and endothelin type a receptor (eta -231 a/g) gene polymorphism. Neurology. 2001;51:1273-1277., , , , , .
- 10Notch 3 mutations in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), a mendelian condition causing stroke and vascular dementia. Ann N Y Acad Sci. 1997;826:213-217., , , et al.
- 17Does a fetal circle of Willis protect against white matter abnormalities in migraineurs? Headache Care. 2006;3:15-20., , , et al.
- 19Headache Classification Committee of the International Headache Society. The International Classification of Headache Disorders, second edition. Cephalalgia. 2004;24(Suppl. 1):1-152.
- 20Measuring nominal scale agreement among many raters. Psychol Bull. 1971;76:378-382..