Brain imaging study identifies two distinct biological migraine subtypes

They've been described as "brain on fire" or "an ice pick through the head." Migraine headaches affect more than one in 10 Americans, and they're so much worse than a regular headache.

As described by those who live with them, migraine headaches are intensely painful and are often accompanied by other debilitating symptoms such as nausea, visual disturbances, and sensitivity to light and noise. These attacks can be incapacitating - migraine headaches are one of the world's leading causes of disability.

Despite their prevalence, migraine headaches are still diagnosed and treated based solely on a patient's description of symptoms. Migraine treatment is pretty much guesswork, said Robert Cowan, MD, a clinical professor of neurology who specializes in headache research.

Recently, Cowan led the largest study to date to categorize migraine patients using a brain imaging technique known as functional MRI. The study, which was published in the journal Cephalalgia on March 26, used fMRI to identify two biological subtypes of migraine. Jaiashre Sridhar, a data analyst at Stanford Medicine, is lead author on the study, and Cowan is senior author.

The new classification could lead to better treatments for the condition, Cowan said, as well as predictions for which patients are likely to develop chronic migraine, a condition defined as having more than 15 days with headaches per month.

Doctors classify migraine as either chronic or episodic (fewer than 15 headache days per month), and current guidelines recommend preventive, daily medication such as beta blockers or anticonvulsants only for chronic migraine patients. Insurance companies often won't cover preventive medication for those who don't meet the criteria for chronic migraine, Cowan said.

But he suspects that many episodic migraine patients could benefit from preventive treatment. The researchers found that patients in one of the two biological subtypes identified in their study had much more severe migraines with more bothersome symptoms but were no more likely to have the chronic migraine diagnosis. The scientists are wondering whether the biological subtype classification will lead to improved treatment decision making, something that is desperately needed.

Right now, treatment decisions for migraine are worse than trial and error. It's darts in the dark."

Robert Cowan, MD, clinical professor of neurology, Stanford Medicine

Data-driven classification

To improve on the chronic-versus-episodic migraine classification system, Cowan and his colleagues wanted to cast the widest net possible. Rather than forming a hypothesis about the biological underpinnings of migraine headaches, they used computational methods to identify subgroups of patients from their imaging data.

"We decided to collect as much data as we could and throw it against the wall - well, the computational wall - and see what sticks," Cowan said.

The study enrolled 111 Stanford migraine patients and 51 control volunteers who don't suffer from migraine headaches. The researchers gathered detailed clinical and demographic information from the participants, then conducted two different types of MRI imaging: structural MRI, which looks at the physical structures of the brain, and fMRI, which looks at the brain's activity based on blood flow between regions.

When they let the computer group their data into clusters, fMRI was more predictive of differences between patients than was the structural imaging. One of the two subtypes, which the scientists call cluster 1, seemed closer to the control group in their brain imaging - this group also had less severe migraine headaches overall. The other subtype, cluster 2, showed big differences in the blood flow between the cortex and subcortical regions of the brain as compared with the other subtype and the control group.

The data shows that patients with the cluster 2 subtype have a different response to sensory input than do non-migraine people and those in cluster 1, Cowan said. It makes evolutionary sense that the brain triggers pain in response to certain sensory inputs - pain makes us want to retreat, and some new-to-us sensations might prove dangerous. But for migraine sufferers, the brain seems to be going over the top, inducing pain in response to daily sensory experiences.

Patients in cluster 2 also had distinct clinical characteristics: They were older, had longer-lasting migraines and were more likely to be disabled by their condition. Overall, cluster 2 seems to have more severe migraines, Cowan said. But interestingly, there was no difference in the frequency of migraine headaches between the two groups, suggesting that the canonical classification of chronic versus episodic migraine might not fully reflect the biology of the disease.

Cowan and his colleagues are now working on migraine classifications based on blood biomarkers and detailed clinical features; they also want to determine whether these subtype classifications can predict treatment response - especially whether someone who doesn't meet the criteria for chronic migraine might benefit from preventive, daily treatment. And because many patients won't be able to get an fMRI due to the method's expense, the scientists are seeking a set of clinical criteria that line up with the biological subgroups.

"It may well be that we could identify someone based on the correlation between certain clinical features and what we know about the imaging cluster and say, 'Here's a patient with five days a month of headache who would really benefit from being on a preventive treatment now,'" Cowan said.

This study was supported by the SunStar Foundation.