Neuroimaging could help separate patients with migraine into two subgroups with distinct clinical and imaging profiles.
In a study, researchers analyzed structural and resting-state functional magnetic resonance imaging (MRI) data from 111 patients with migraine and 51 controls. The migraine cohort included 75 patients with chronic migraine and 36 with episodic migraine who were classified according to International Classification of Headache Disorders, 3rd edition criteria. Using cortical thickness, cortical volume, subcortical volume, and functional connectivity measures across cortical and subcortical regions, the researchers applied principal component analysis followed by hierarchical agglomerative clustering to identify multimodal imaging-derived migraine subgroups.
Using the multimodal MRI model, the researchers identified two clusters: Migraine Cluster 1 (M1f + s) with 48 participants and Migraine Cluster 2 (M2f + s) with 63 participants. Patients in the M2f + s cluster were older and had longer migraine duration, greater migraine-related disability, and lower pain self-efficacy compared with patients in the M1f + s cluster. The rates of medication overuse headache, aura, prodrome, and allodynia were also higher in the M2f + s cluster, although headache frequency and intensity were comparable between the groups.
Functional connectivity analyses showed that those in the M2f + s cluster had broadly increased connectivity between the dorsal attention, somatomotor, and visual networks and multiple subcortical regions, including the thalamus, basal ganglia, hippocampus, and amygdala. Compared with controls, patients in the M2f + s cluster demonstrated increased cortical-subcortical connectivity, whereas M1f + s showed no statistically significant functional connectivity differences relative to controls.
Structural analyses showed lower cortical volumes in the M2f + s cluster across frontal, parietal, temporal, and insular regions compared with the M1f + s cluster. The researchers also identified a small but significant reduction in pars orbitalis thickness in the M2f + s cluster. No statistically significant subcortical volumetric differences were observed between the clusters, and neither subgroup demonstrated structural differences compared with controls.
Secondary unimodal clustering analyses compared multimodal clustering with functional-only and structural-only models. Functional-only clustering showed moderate agreement with the multimodal solution, with an adjusted Rand index of 0.427, suggesting that functional connectivity contributed substantially to subgroup differentiation. Structural-only clustering showed minimal overlap with the multimodal clusters, with an adjusted Rand index of 0.001, which the researchers said reflected “an orthogonal dimension of heterogeneity captured by structural variation.”
The researchers also performed a sensitivity analysis using finer functional parcellation. The resulting clustering solution remained broadly consistent with the primary multimodal model, although reduced signal coverage lowered the usable sample size.
The researchers noted several limitations, including the modest sample size and cross-sectional design, which limited assessment of subgroup stability over time. Migraine phase during imaging and preventive medication use were not systematically recorded, which may have introduced variability. The study also was not designed to compare imaging-based subgrouping with clinical-only subgrouping approaches.
“In conclusion, this study applied a data-driven integration of functional and structural MRI features to characterize heterogeneity within migraine,” wrote lead study author Jaiashre Sridhar, of the Department of Neurology & Neurological Sciences at Stanford University, and colleagues. “Together, these findings suggest that migraine encompasses distinguishable patterns of functional and structural brain organization.”
The study was supported by the SunStar Foundation. Co–study author Robert P. Cowan disclosed advisory relationships unrelated to the study with Alder, Amgen, Allergan, Biohaven, Curex, Teva, and Xoc. The study authors reported no other conflicts of interest.
Source: Cephalalgia
