For this paper, I decided to complete two complementary reviews. This one for general readers can be considered a background and a summary for the Journal Club Scientific Review.
For some time there has been a nagging concern among clinicians that migraine is associated with premature vascular changes in the brain. Given how common migraine is, and how commonly imaging is performed as a screening investigation for headache, there arises all too commonly an awkward situation where imaging is performed in a patient with migraine to rule out sinister pathology, and then the imaging is “not quite normal”. In fact the imaging indicates the presence of vascular changes that are typically present mainly in older people. Hardy reassuring.
Does this mean that every migraine attack is causing a mini-stroke, or that migraineurs, when they grow older, are more susceptible to stroke or to vascular dementia or to pre-frontal gait and balance problems? How aggressively should we address vascular risk factors in all migraine patients, about 12% of the adult population? Should we perform MRI scans on all 12%, and address risk factors in the sizeable proportion with the excess lesions, or address risk factors in all, or in none? Should we be thinking in terms of secondary prevention measures, rather than primary prevention? (Secondary prevention means preventing stroke or heart disease when such events have already occurred. The balance of risks is consequently shifted in favour of intervention despite potential side effects or risks.) What about echocardiography to screen all migraineurs for cardiac sources of emboli and for mitral valve prolapse? What about a bubble study to investigate patent foramen ovale? The questions multiply and the answers are frustratingly lacking.
These concerns over MRI appearances were confirmed by epidemiological findings, including the CAMERA study (Cerebral Abnormalities in Migraine – an Epidemiological Risk Analysis). In nearly 300 subjects with migraine, the female subgroup was indeed found to have an excess of small scattered white matter changes on MR imaging compared to 140 age, sex and other risk factor matched controls. Furthermore, the more frequent the migraines, the greater the number of lesions, indicating that there could be some cumulative lesioning effect from migraine attacks.
However, this study merely corroborated the imaging findings. It did not indicate whether or not they actually mean anything for patients. Therefore the CAMERA study followed up its patients, measuring changes in lesion load and recording cognitive ability by IQ tests; the findings after 9 years are presented in CAMERA 2, the subject for this review.
Around two-thirds of the original CAMERA 1 study subjects with and without migraine were followed up. In females, it was found that 77% of patients with migraine had worsening of a certain pattern of imaging abnormalities called deep hemispheric white matter lesions, compared with 60% of female controls. One expects some progression simply due to age, the mean age by this second study being 57 years. The more prevalent progression in the migraine patients was nevertheless statistically significant (p=0.04). Progression of other types of brain lesions was not significantly different between female migraineurs and controls, nor was there a migrainous association in men with any kind of MRI lesion or progression thereof. Unlike the baseline findings from CAMERA 1, further progression in the number of white matter lesions was not associated with a higher frequency of migraine attacks.
Most importantly, the study failed to find any relationship between presence or absence of MRI lesions and cognition. However, overall I would personally take these finding as leaving me “a little less worried than I was before” rather than “reassured”. This is because of the statistical detail.
The authors chose to analyse the cognitive (and fine movement task) data by lumping all the migraine and non-migraine patients together and then dividing them into the worst fifth regarding lesion load and the best four fifths. Using a statistical model involving linear regression, they found that, after correcting for prior educational level, age and sex, there was a trend for worse cognition in the smaller high lesion load group compared to the larger low lesion load group but this did not reach significance (p=0.07).
However there is a difference between saying, “the lack of statistical significance means there is no evidence for an effect of lesion load on cognition”, and “the lack of statistical significance means there is positive evidence for no effect of lesion load on cognition”. This difference is often lost on journalist and publicists. Although the statistics cannot prove that cognition is worse with higher lesion load, with that p-value I for one would like to be in the low lesion load group!
They then analysed the all-important migraine issue by bringing whether or not the subjects had migraine into the high vs low lesion load cognition statistical model and they found that having migraine did not influence this (said to be lack of) effect (p=0.3). But if the effect is really borderline rather than absent, might the migraine influence be too?
There was also the very clearly non-significant statistic (p=0.9) that the migraine patients overall had cognitive scores that were not significantly worse than non-migrainous controls, which is reassuring though I think this was a straight comparison rather than correcting for possibly higher original cognition or possibly higher educational level.
Finally, high lesion load in the CAMERA 1 study 9 years earlier did not predict worsened cognition at the time of CAMERA 2. In other words, it seems to be more the age-related subsequent accumulation of lesions that possibly matches with poor cognition rather than the original migraine associated lesions. (Remember, nearly as many non-migrainous patients had progression in white matter changes over the nine years as migraine patients.)
While these two latter points are somewhat reassuring, we still do not get a clear answer to the question, “In the subset of female migraine patients with high lesion load, did their cognition deteriorate more from nine years earlier than that of the migraine patients with low lesion load or that of the controls with low lesion load?”
Returning to the original clinical scenario, given all the whys and wherefores I don’ t think we can draw any firm conclusions from this study to provide reassurance to patients with migraine. Yes, migraineurs have progression in lesions more than expected for age. No, these are not associated with ongoing frequency of migraine attacks and no they are not found to be associated with impaired cognition nine years later.
One must also place studies in their context. Reviewing this paper prompted me to look further into the literature. In fact there is a reasonable body of recent evidence from long-term follow-up of migraine patients in general that there is no progressive cognitive impairment. This therefore provides further support for the argument that the MRI lesions seen in migraine do not have this clinical significance.
Nevertheless, I still cannot be confident that in no migraine patient is there any significance to their lesion load, beyond that associated with other coincidental risk factors such as diabetes. I think further follow-up of this study cohort would be helpful. For example, another ten years later when the subjects will on average be in their sixties, will there be any greater deterioration in the already-measured cognitive scores in the subset of migraine patients with more highly progressive lesions than in non-migraine patients with more highly progressive lesions? More importantly, are the high lesion load patients with migraine becoming clinically demented, or suffering increased strokes or progressive gait impairment?
I can only say that, working retrospectively from my own clinical experience, an excess risk of stroke and other vascular diseases is not something I have particularly observed in patients who had migraine when they were younger, unlike the situation in cigarette smokers and diabetics. On the other hand, in the elderly population, the occurrence of migraine attacks does seem to be a marker of vascular disease. Perhaps it is the age of the patient with migraine that is the key, and the slightly mixed findings of the study reflect that they have selected a rather mixed-aged cohort.
Link to Scientific Review of this topic.