Insular functional connectivity in migraine with aura

Migraine is a disabling disorder affecting up to 15% of the global population, and is the second leading cause of years lived with disability [1, 2]. Migraine pain is related to the involvement of the trigeminovascular system [3], but migraine is not only a pain, as it is also accompanied by several sensory, autonomic, affective and cognitive disorders. These symptoms are posited to result from multiple brain networks involvement within the brainstem, the subcortical and cortical areas, beyond the trigemino-vascular system [4]. In about one third of migraine patients, migraine attacks are accompanied by an aura which is a transient progressive and fully reversible central neurological symptom, most often visual, occuring before the headache. The insula is involved in multiple cerebral functions such as sensorymotor processing, pain, taste, interoception, autonomic control, emotions, attention or salience which refers to the ability to select the most relevant information among multiple internal and external stimuli [5]. Parcellation of insula resulted in two architectonic subdivisions (posterior granular area and anterior dysgranular area) to thirteen multi-modal MRI subdivisions [5, 6]. Nevertheless, data-driven meta-analysis of human functional imaging studies supported a tripartite subdivision of the insula into a ventroanterior area, a dorsoanterior area and a posterior area. The ventroanterior insula is functionally coupled with limbic areas and is associated with emotion, chemosensation and autonomic functions. The dorsoanterior insula is connected to the anterior cingulate cortex and the dorsolateral prefrontal cortex and plays a role in cognitive tasks and executive control. Conversely, the posterior insula is connected to the somatosensory cortex and the suplementary motor area resulting in pain and somatosensory functions involvement [7]. Evidences support the involvement of the insula in several features of migraine pathophysiology from the ictal phase to chronicisation. During the ictal phase of spontaneous migraine attacks, Positron Emission Tomography (PET) studies revealed activation of bilateral insula cortex as well as other cortical areas, brainstem and diencephalic nuclei [8, 9]. In addition, functional MRI studies during the ictal phase showed a stronger activation of the anterior insula in response to olfactory stimulations, but a decreased functional connectivity (FC) of the anterior insula with the medial prefrontal cortex within the Default Mode Network (DMN) inversely proportional to the pain intensity [10, 11]. Another study showed a higher FC between the right thalamus and the left insular cortex during spontaneous migraine attacks [12]. During the intercital phase of migraine without aura (MO), the right posterior insula was identified as a hub of FC more strongly connected to the supplementary motor cortex and the paracentral lobule among other brain areas [13]. In high frequency migraine, defined by 8 to 14 monthly migraine days, compared to low frequency migraine, heat painful stimulations of the hand induced lower controlateral anterior insula and bilateral inferior insula activations, but a higher connectivity of bilateral insula with the left post central gyrus [14]. In chronic migraine, number of years of chronic migraine were correlated to the resting state FC between bilateral anterior insula and the right mediodorsal thalamus, as well as to the FC between the right anterior insula and the periaqueductal grey matter (PAG) [15]. Overall, the insula is posited to play a key role in migraine, acting as a « hub» of integration of autonomic, sensory, affective and cognitive functions [16]. Insula in migraine with aura (MA) is of specific interest as previous studies have found specific alterations of insular connectivity in MA. The anterior insula had a reduced connectivity with occipital areas in MA compared to MO and Healthy Controls (HC), and the connectivity changes between the left anterior insula and occipital areas were negatively correlated with headache severity in MA only [17]. In a study investigating cognitive functions in migraine and assessing the DMN, patients with MA presented an increased FC between the right insular cortex, the left angular gyrus, the left supramarginal gyrus, the right precentral gyrus and the right postcentral gyrus compared to MO. In patients with complex MA, defined by more than visual symptoms, the right anterior insula was more strongly connected within the sensorimotor network compared to simple visual aura and MO, and this increased FC could discriminate between complex MA and simple visual aura [18]. Moreover, in a PET/MRI brain study, uptake of [11C]PBR28, a glial activation maker, in the right posterior insula was correlated to the number of MA attacks [19]. However, this result was not compared to MO.These observations suggested that the insula exhibited altered connectivity in MA, however these studies have not taken into account the functional division of the insula. In fact, the studies have focused either on the salience network, which includes the anterior insula [17, 20, 21], or on the somato-sensory network, which includes the posterior insula [18] or to a few regions of interest that did not explore the insula in its subdivisions. To our knowledge, the FC of the insula’s functional subdivisions has not yet been comprehensively studied in MA. Therefore, in the present study, we aimed at investigating the bilateral insular connectivity in MA using seeds in the anterior, posterior, dorsal, middle and ventral insula.

This explorative study aimed at investigating whether insula exhibits differences in FC in patients with MA using a comprehensive seed-to-voxel analysis of right and left insula with six seeds located within the anterior, posterior, dorsal, middle and ventral insula. We found an increase connectivity of both right and left antero-dorsal insula with the cerebellar vermis VI in patients with MA. This increased FC was not found in patients with MO. Previous studies in migraine have shown that insula exhibits alterations of FC with brain structures involved in pain processing such as thalamus [12], PAG [26, 27], somatosentory cortex [18], or cognitive function such as the default mode network [21, 28] or salience network [17, 20, 21]. In MA specifically, insula exhibited altered FC with occipital cortex and somatosensoriel cortex only in complex MA [17, 18]. However, these findings relied mostly on investigations of other brain areas or partial explorations of insula. To the best of our knowledge, our result is new and its originality could be explained by the methods which consisted of a comprehensive insular FC analysis by distinguishing insula subdivions from each other. Other methodological differences with previous studies should be considered. Previous studies showing an increased FC of the insula with the thalamus involved MO during spontaneous migraine attacks [12], an increased connectivity of insula with the PAG involved patients with allodynia [26, 27], an increased connectivity of insula with the somatosensory cortex involved patients with complex auras [18]. One study showing a lower correlation with the visual cortex set a less conservative threshold of significance for statistical maps [17]. Our study showing an increased FC of both right and left antero-dorsal insula with the cerebellar vermis VI provides a new perpective on the role of insula in MA. Because our study did not demonstrate a correlation between this FC and clinical features of MA, we acknowledge that the clinical significance of our finding is currently undetermined. However, knowledge on the function of the insula and the vermis VI allowed us to postulate two hypotheses to explain this increased FC: one related to the central integration of pain, the other to the control of the parasympathetic autonomic nervous system. Animals and humans studies have highlighted the involvement of the cerebellum in pain [29]. Indeed, studies in healthy volunteers, using fMRI showed an activation of cerebellar lobule VI, VIIIa, crus I and vermal lobule VIIIa evoked by painful stimulation of the left nostril. The cerebellum presented an increased FC with structures involved in pain processing such as rostral pons, PAG, thalamus and cortices regions including insula and face area in the precentral gyrus [30]. Compared to HC, migraine patients presented a higher activation of PAG and left cerebellum crus I in response to nociceptive trigeminal stimulations. Moreover the vermis VI belonged to a cluster that comodulated with migraine-phase [31]. Moreover, functional MR studies revealed cerebellar activation evoked by heat painful stimulations in migraine [32‐34]. Although these evidences stressed the role of cerebellum in trigeminal pain processing in migraine, our result did not support clearly a role of vermis-insula FC in pain because this FC was not correlated to migraine feature and was not found in patients with MO. The insula is part of the autonomic nervous system (ANS) network, as well as the anterior cingulate cortex, the pre-frontal cortex and the amygdala [35]. Some studies have suggested that the cerebellum is also involved in the regulation of the cardiovascular ANS. Functional neuroimaging studies in human confirmed that cerebellum is activated during tasks challenging cardiovascular ANS and both anterior insula and vermis cerebellum seem to be involved in the regulation of parasympathetic tone [36‐38]. A meta-analysis published in 2013 included studies analysing peripheral signals in response to ANS stimulation by cognitive, affective and somatosensory autonomic nervous system tasks in conjunction with brain imaging in healthy subjects. The ANS response was classified as sympathetic or parasympathetic based on measures of heart rate variability and electrodermal activity. This meta-analysis showed parasympathetic activation of the dorsoanterior insula and vermis VI in addition to other brain structures such as the amygdala or posterior cingulate cortex [38]. The shared role of the cerebellum and insula in the regulation of the cardiovascular ANS is also supported by two activation studies and a connectivity study during ANS stimulation tasks: in a PET study, Critchley et al. have explored the cerebral activation during isometric exercise and mental arithmetic stressor tasks. These tasks induced variation in mean arterial blood pressure (MAP) and heart rate (HR) and were accompanied by an activation of the midline cerebellum, the brainstem in the region of the pontine reticular nuclei and the right dorsal cingulate cortex. In a conjunction analysis, the activation of both the cerebellar vermis and the insular cortex covariated with MAP and HR [39]. Baker et al., in an fMRI study, showed that lower-body negative pressure maneuver (LBNP) induced activation of bilateral insula. The cerebellum and the bilateral insula were activated during the recovery phase of LBNP [40]. A further analysis revealed that vermis VI was part of a cerebellar connectome functionally coupled with bilateral insula, as well as anterior cingulate cortex, bilateral thalamus and bilateral putamen [41]. These previous observations supported the hypothesis that the increased FC between dorsoanterior insula and vermis VI in MA could be related to the cardiovascular parasympathetic ANS control. However, our study was unable to confirm this hypothesis as no recording of cardiovascular parameters was performed during the MR scan. Nevertheless, this hypothesis is likely given the cardiovascular autonomic disorders observed in MA. It has previously been shown that both MA and MO are at increased risk of syncope and orthostatic intolerance [42]. However, one study found that only MA was associated with increased risk of syncope after adjustment on confounders [43]. A review of studies assessing cardiovascular autonomic balance in migraine showed a trend toward greater autonomic dysfunction in MA than in MO, with sympathetic dysfunction being more common than parasympathetic dysfunction [44]. These previous observations supported the hypothesis of ANS-related alteration of FC between dorsoanterior insula and vermis VI in MA. The strengths of our study were the comprehensive exploration of the insula in a seed-to-voxel analysis. The result remained significant after correction for multiple comparisons and was found bilaterally. A random result seemed thus unlikely. ROI-to-ROI analysis excluded the involvement of the lingual cortex in the vicinity and confirmed the functional coupling of the vermis VI with the antero-dorsal insula. Some limitations should be mentioned. The study involved a small sample of volunteers, which may have underestimated differences in connectivity with other brain areas. Although a previous study suggested gender differences of insular connectivity in pain [45], our study sample did not allow for testing gender differences of insular connectivity in migraine. Patients with MO were included from another study previously conducted at our center but the MRI protocol was the same as in the study including patients with MA and HC. Finally, assessment of the cardiovascular ANS during fMRI was not performed which mitigate the interpretation of our result. Further studies are warranted to determine whether the increased functional coupling of antero-dorsal insula with vermis VI reflected an increased parasympathetic tone as we may hypothesize.

In MA, the bilateral antero-dorsal insula was strongly functionally coupled with the cerebellar vermis IV. Because both regions are involved in the control of the parasympathetic cardiovascular ANS, this functional connectivity could reflect the cardiovascular features of MA. Further research is needed to explore this hypothesis.