Dec. 29, 2017
Mayo Clinic researchers are using brain MRI to develop objective biomarkers for the classification of migraine. The existing, subjective criteria are based on the expert opinions of contributors to the International Classification of Headache Disorders (ICHD). The Mayo Clinic researchers hope to validate or refine those criteria and perhaps identify new migraine subtypes, paving the way for better targeted therapies.
"We currently have limited ability to practice precision medicine for patients with migraine," says Todd J. Schwedt, M.D., a consultant in Neurology at Mayo Clinic in Phoenix/Scottsdale, Arizona, and a member of the committee that helps compile the ICHD. "Only about 40 to 45 percent of patients respond to any one of the first line migraine-preventive therapies, and we are not able to accurately predict which patient will respond to which therapy. I think there's a reason for that — our current classification does not identify all the heterogeneity that exists among groups of patients with migraine."
The MRI studies led by Dr. Schwedt are currently used only for research. However, identification of new patient subgroups from these studies might eventually be used in the clinical domain. "I'm a firm believer that there are additional subtypes of migraine beyond those that are commonly recognized," Dr. Schwedt says. "Imaging is one way we might identify these subgroups."
Structural and functional MRI
Functional MRIs of brain migraineurs and healthy controls
Functional MRIs show brain activations in migraineurs and healthy controls in response to moderately intense heat pain applied to the arm. Image reprinted with permission from Cephalagia.
Brain region characteristics help differentiate chronic and episodic migraine
Illustration shows brain regions in which surface area, thickness or volume measures contributed to a classifier differentiating patients with chronic migraine and episodic migraine. Brain measurements were obtained using structural MRI. Illustration reprinted with permission from Headache.
The Mayo Clinic research uses both structural and functional MRI of people with migraine and other types of headaches, and healthy controls. Structural MRIs evaluate factors such as the volume of various brain regions, cortical thickness, surface area, curvature of the brain and brain shape. Diffusion tensor imaging is used to assess white matter integrity.
Functional MRI involves both resting state functional connectivity, to assess how areas of the brain are connected and communicate, and event-related functional MRI to determine how participants' brains respond to a painful stimulus.
"So far, we've built classification models for both episodic and chronic migraines, using structural as well as functional data and a combination of the two," Dr. Schwedt says.
In a study published in the August 2017 issue of Cephalagia, Dr. Schwedt and colleagues used machine-learning techniques and data from resting-state functional MRI of pain-processing regions to develop biomarkers that distinguish between individuals with migraine and healthy controls. Six brain regions — the bilateral amygdala, right middle temporal, posterior insula, middle cingulate and left ventromedial prefrontal — had the most discriminative power.
The researchers were able to classify individual brain MRIs as belonging to a person with migraine or a healthy control with an overall accuracy of 81 percent and a best accuracy of 86 percent. Migraineurs with longer disease durations were classified more accurately than those with shorter disease durations.
"We can look at the functional MRI and tell you with greater than 80 percent accuracy whether that MRI belongs to somebody who has chronic versus episodic migraine, or chronic migraine versus a healthy control," Dr. Schwedt says.
In a subsequent study published in the July 2017 edition of Headache, structural MRIs measured regional cortical thickness, volumes and cortical surface areas in the brains of migraineurs and healthy controls. Automatic data-driven analysis clustered the MRIs into two subgroups. People with migraine in the first subgroup had more severe allodynia symptoms during migraine attacks, more years with migraine and higher Migraine Disability Assessment scores. Headache frequency and aura status weren't significantly different between the two subgroups.
"Allodynia occurs in the majority of people during a migraine attack. It's been suggested that allodynia might affect migraine treatment response and disease prognosis, and our study suggests that it affects brain structure," Dr. Schwedt notes. "The presence or severity of allodynia could be considered when defining migraine subgroups."
The imaging studies may also shed light on whether anomalies in brain structure or function are present at birth or result from migraine attacks. "We have found that the more severe a person's migraines, the greater accuracy we have classifying that person according to brain MRIs. That probably means that the brain changes occur secondary to recurrent attacks," Dr. Schwedt says. "But it's possible that certain brain structures or functions at baseline predispose a person to migraine. We need large, longitudinal studies to determine directionality."
Another focus of research compares MRIs of people with migraine and people with post-traumatic headaches. Although symptoms are often similar or identical for both types of headache, the researchers have found differences in brain structure between the two groups. "That suggests there might be differences in the underlying pathophysiology of migraine and post-traumatic headache," Dr. Schwedt says.
"Our ultimate goal is to identify objective differences among people with migraine and other headache types that predict treatment responses and allow us to provide more-targeted therapies for these patients," he says.
For more information
Chong CD, et al. Migraine classification using magnetic resonance imaging resting-state functional connectivity data. Cephalalgia. 2017;37:828.
Schwedt TJ, et al. Migraine subclassification via a data-driven automated approach using multimodality factor mixture modeling of brain structure measurements. Headache. 2017;57:1051.