Human Studies

Cluster headache (CH) shows a more severe clinical picture than migraine (Mig). We tested whether brain changes can explain such difference. Multimodal MRI was acquired in attack-free patients with CH (n = 12), Mig (n = 13) and in normal controls (NC, n = 13). We used FSL for MRI data analysis and nonparametric permutation testing for voxelwise analyses (p < 0.01, corrected). CH showed lower grey matter (GM) volume, compared to Mig and NC, in frontal cortex regions (inferior frontal gyrus and frontal pole [FP], respectively) and, only compared to Mig, in lateral occipital cortex (LOC). Functional connectivity (FC) of CH was higher than Mig and NC within working memory and executive control networks and, only compared to Mig, between cerebellar and auditory language comprehension networks. In the attack-free state, the CH brain seems to be characterized by: (i) GM volume decrease, compared to both Mig and NC, in pain modulation regions (FP) and, only with respect to Mig, in a region of visual processing modulation during pain and working memory (LOC); (ii) increased FC at short range compared to both Mig and NC and at long range only with respect to Mig, in key cognitive networks, likely due to maladaptation towards more severe pain experience.

ICHD-3 criteria for chronic migraine (CM) include a mixture of migraine and tension-type-like headaches and do not account for patients who have a high frequency of migraine but no other headaches.

The definition of chronic migraine has long been debated. Recently, it was suggested to define subjects with at least 8/migraine days as chronic migraine; that is, incorporating so-called high frequency episodic migraine (eight or more migraine days but less than 15 headache days per month).

This study aims to investigate whether intra-network dynamic functional connectivity and causal interactions of the salience network is altered in the interictal term of migraine. 32 healthy controls, 37 migraineurs without aura and 20 migraineurs with aura were recruited. Participants underwent a T1-weighted scan and resting-state fMRI protocol inside a 1.5T MR scanner. We obtained average spatial maps of resting-state networks using group independent component analysis, which yielded subject-specific time series via a dual regression approach. Salience network ROIs (bilateral insulae and prefrontal cortices, dorsal anterior cingulate cortex) were obtained from the group average map via cluster-based thresholding. To describe intra-network connectivity, average and dynamic conditional correlation was calculated. Causal interactions between the default-mode, dorsal attention and salience network were characterised by spectral Granger's causality. Time-averaged correlation was lower between the right insula and prefrontal cortex in migraine without aura vs. with aura and healthy controls (p<0.038, p<0.037). Variance of dynamic conditional correlation was higher in migraine with aura vs. healthy controls and migraine with aura vs. without aura between the right insula and dorsal anterior cingulate cortex (p<0.011, p<0.026), and in migraine with aura vs. healthy controls between the dorsal anterior cingulate and left prefrontal cortex (p<0.021). Causality was weaker in the <0.05 Hz frequency range between the salience and dorsal attention networks in migraine with aura (p<0.032). Overall, migraineurs with aura exhibit more fluctuating connections in the salience network, which also affect network interactions, and could be connected to altered cortical excitability and increased sensory gain.