Our findings demonstrate an interictal increase in resting-state intrinsic connectivity between the PAG and both nociceptive and sensory processing pathways in migraineurs relative to age and gender- matched controls. Connectivity in some of these pathways was stronger as the monthly frequency of migraine attacks increased. Conversely, the greater the number of attacks was, the lower the functional resting-state connectivity appeared to be between the PAG and several brain regions with a predominant role in pain modulation (prefrontal cortex, anterior cingulate, amygdala). Since the occurrence of the next migraine attack was not recorded, whether the observed association between migraine frequency and changes in PAG connectivity reflects either an adaptive mechanism or an abnormal status in cortical excitability that predisposes to a migraine attack needs to be further clarified. Excitability in the visual cortex has been shown to increase interictally in pediatric MWoA, although it tends to decrease or normalize in the peri-ictal period23
Previous neuroimaging findings indicate that the brainstem covers a crucial role in acute migraine24
, and positron emission tomography (PET) studies report metabolic changes in either the dorsal rostral brainstem or in the locus coeruleus and dorsal raphe nucleus of migraineurs25, 26
. Dysfunction of the regulation of the brainstem pain-inhibiting circuitry, of which the PAG is best known, may provide an explanation for many of the facets of migraine headache.
Compared to healthy subjects, migraineurs showed, in the absence of any pain, significantly greater intrinsic connectivity between PAG in several brain regions including thalamus, posterior parietal cortex (angular and supramarginal gyri), anterior insula, S1, M1, and partly S2. While S1, S2, anterior insula, and thalamus are part of the fundamental core network found active by several neuroimaging studies in different pain conditions activity in other regions including the posterior parietal cortex and M1 has also been reported in nociception27
. These findings could reflect a condition of hyperexcitability of pain pathways within the central nervous system (CNS), which is thought to represent a crucial event in the pathophysiology of migraine. Transcranial magnetic stimulation (TMS), and magneto-encephalography (MEG) studies have shown generalized interictal changes of excitability in the cerebral cortex in migraine with alteration of excitability in the motor and visual cortices 28-33
. In addition, in patients with unilateral migraine, in a headache-free condition, diffuse hypersensitivity of the peripheral nerves has been measured, leading to further evidence to the presence of a state of hyperexcitability of the CNS of migraineurs34
. Interestingly, central hyperexcitability does not seem to be a feature of chronic either tension type or cervicogenic headaches, which are characterized by different pathogenetic and clinical features than migraine35
A systematic investigation of intrinsic connectivity patterns of the PAG by resting-state fMRI in a large cohort of healthy subjects reports that PAG activity is positively correlated with surrounding subcortical brain regions including midbrain tegmentum, substantia nigra, raphe nucleus, thalamus, and hypothalamus, as well as with cortical regions including the anterior cingulate, and anterior insula36
. In our study, in some of these areas (anterior insula, NCF in the midbrain, and hypothalamus) the strength of intrinsic connectivity with PAG positively correlated with disease severity as measured by the number of migraine episodes per month. Additional areas involved in nociception and somatosensory processing including S2 and the posterior parietal cortex also showed increased resting-state connectivity with PAG with increasing in the monthly frequency of migraine attacks.
The finding that hypothalamic intrinsic correlations with PAG are related to the frequency of migraine attacks extends previous PET data that have first described activation of the hypothalamus in migraineurs during the headache phase, which persisted after headache remission with sumatriptan37
. While specific hypothalamic activations have been consistently observed in cluster headache, and linked to the autonomic manifestations of trigeminal autonomic cephalalgias, the role of the hypothalamus in migraine has not been clarified yet24
. The hypothalamus is integrated in the hypothalamic-pituitary-adrenal (HPA) axis, which is also involved in mood disorders such as stress and anxiety. In addition to pain modulation, the PAG is well known to participate in mood and emotion regulation, which may be carried out through a pathway involving the HPA axis also36
. In this context, although subjects with depression were excluded from our study, an increase in the intrinsic correlations between PAG and hypothalamus in relation to the frequency of migraine attacks can be viewed as a part of a stress/anxiety response of the brain to worsening of the disease. Behavioral data assessing anxiety levels in patients with migraine may help to clarify this aspect.
In migraineurs, we also found that the greater the frequency of migraine episodes was, the lower the resting-state functional connectivity appeared to be between PAG and several brain foci, mostly located in the bilateral prefrontal (medial, dorsomedial) cortex, but also in the anterior cingulate and amygdala, both involved in descending pain modulation and nociception27
, and in somatosensory processing regions including the angular gyrus (posterior parietal cortex), right S1, and M1. The pattern, however, was clearly dominated by a widespread reduced connectivity between PAG and prefrontal cortex that correlated with the increase in migraine attacks frequency. Pain studies have emphasized the role of prefrontal areas in controlling the functional interactions among key nociceptive brain regions in order to modify the perceptual correlates of pain, specifically by driving endogenous pain-inhibitory mechanisms27, 38
. Our findings could therefore indicate an interictal dysfunction of the descending inhibitory system that in turns contributes to the development of migraines. TMS and PET imaging demonstrated that in patients with chronic migraine increase in visual cortical excitability is accompanied by brainstem activation and inhibition of specific cortical areas including the medial frontal and parietal cortex, and somatosensory area, suggesting a potential dysfunction of inhibitory pathways39
. Interestingly, this pattern of significantly reduced connectivity between PAG, prefrontal cortex, and anterior cingulate was also observed in migraineurs with allodynia compared to migraineurs without allodynia, corroborating the hypothesis that pain modulatory neurons might be involved in the development of allodynia.
Taken together, our data demonstrate the presence of interictal dysfunctional dynamics within PAG networks in episodic migraine. Since resting brain activity is either increased or decreased within multiple networks in other chronic pain conditions including fibromyalgia40
and diabethic neuropatic pain41
, future studies may clarify whether the changes that we observed within the PAG networks can be interpreted as a “brain signature” of migraine; or rather they are shared by other chronic pain conditions.