Are Those White Matter Lesions in Migraine Real?

Abstract & Commentary

Source: Kruit MC, et al. Migraine as a risk factor for subclinical brain lesions. JAMA. 2004;291:427-434.

The possible association of migraine and stroke remains an active area of controversy in clinical neurology. Since migraines predominantly affect young people and the medications used can cause vasoconstriction, assessing the full vascular risk profile of migraine is important from a diagnostic and therapeutic point of view. To date, the literature is incomplete. Perhaps the best study on the subject is a case-control one from Tzourio and associates1 suggesting that for women younger than 45, migraine with aura is only a minor stroke risk factor in association with smoking and oral contraceptive use. Adding to the complexity of the issue is presence of nonspecific white matter lesions (WML) on MRI scans of migraineurs. The reports of these findings include relatively small numbers and shed little light on the overall prevalence in migraine patients selected from the general population. Kruit and colleagues report on the results of a large Dutch population-based MRI study to look at whether migraine is associated with WML on MRI and if so, what migraine subtypes in particular, as well what regions of the brain, might be more vulnerable.

A total of 435 patients were selected for MRI scans and were subgrouped by IHS criteria accordingly: migraine with aura (n = 134); migraine without aura (n = 161); and age-matched headache-free controls (n = 140). No patients had a history of stroke or TIA. The mean age was 48, and 75% were female. Migraineurs and controls were well matched for cerebrovascular risk factors including blood pressure, smoking, diabetes, and cholesterol. Medication usage such as triptan and ergotamine were recorded as covariables. A single-blinded neuroradiologist reviewed all the MRI scans and rated the infarcts and WMLs recording number, size, and location. WMLs had to be hyperintense on all sequences and were scored with semi-quantitative measures with validated scales. WMLs were differentiated into deep WML (DWML) and periventricular WML (PVWML).

Sixty infarcts ranging in size from 2 mm to 21 mm were recorded from 31 patients. There was no difference in infarct rate between control and migraine (5% vs 8%; P = .23). When considering location, the migraineurs had a 7 times higher risk of posterior circulation infarct predominantly cerebellum (5.4% vs 0.7%; P = .02). This was largely driven by the higher prevalence of posterior circulation infarct in the migraine with aura group compared to the migraine without aura (2.2% vs 8.1%; P = .03). With respect to nonspecific WML, there was no difference in the distributions and severity grade between controls and migraine. Thirty-eight percent of both groups had at least 1 medium-sized WML. There was no association in WML of any kind between men and migraine. However, there was increased risk of only DWML (not PVWML) in women with migraine (odds ratio, 2.1). The risk did not vary in "with aura" or "without aura" subtypes but correlated best to migraine frequency—with more than 1 attack per month, the odds ratio was 2.6. The density of WMLs was not affected by cerebrovascular risk factors or class of medication use.


The study raises several interesting questions, and the methodology improves upon the selection biases that have compromised similar previous studies of this kind. However, as a population-based essentially retrospective study, Kruit et al should be reluctant to make statements of causality. All we now know is that in this cohort migraine with aura patients have a higher prevalence of predominantly cerebellar "infarcts," and this risk increases with increasing headache attacks. Also women with migraine have a higher prevalence of deep WMLs, and this also increases with increasing headache frequency. It is important to remember that these MRI findings were clinically silent (aside from the migraine, of course). In fact, Kruit et al’s designation of the MRI changes as indeed "infarcts" might be presumptive. What we still don’t know from this study is how to think of the ischemic stroke risk in migraine patients and, in general, whether this sheds further light onto the pathophysiology of migraine. Certainly, the predominance of cerebellar MRI changes is notable given the association of cerebellar degeneration with the CANCNA1A mutation in the rare autosomal dominant familial hemiplegic migraine subtype. But the question as to whether the "infarct" changes or the WML are indeed primarily a vascular occlusive ischemic phenomenon or represent a direct neuronal apoptotic event is still unclear. The fact that there was no association with cerebrovascular risk factors argues against a vascular occlusive phenomenon. After all, this has always been the central question of migraine—vascular or neuronal. The current study helps the clinician think about his/her migraine patient who comes back from the MRI with an "abnormal" scan but does not reveal any more deeper secrets into the migraine process. — Jeffrey Reich

Dr. Reich, Assistant Professor of Neurology, New York Presbyterian Hospital- Cornell Campus, is Assistant Editor of Neurology Alert.


1. Tzourio C, et al. BMJ. 1995;310:830-833.