Altered structure of the vestibular cortex in patients with vestibular migraine.

Abstract Introduction Previous voxel‐based morphometry (VBM) studies have revealed changes in brain structure in patients with vestibular migraine (VM); these findings have improved the present understanding of pathophysiology. Few other studies have assessed the association between structural changes and the severity of dizziness in VM. This study aimed to examine the structural changes and cortical morphometric features associated with migraine and vertigo attacks in patients with VM. Methods Twenty patients with VM and 20 healthy normal volunteers were scanned on a 3‐tesla MRI scanner. The gray matter volume (GMV) was estimated using the automated Computational Anatomy Toolbox (CAT12). The relationship between clinical parameters and morphometric abnormalities was also analyzed in VM. Results Compared with controls, VM patients have decreased GMV in the prefrontal cortex (PFC), posterior insula–operculum regions, inferior parietal gyrus, and supramarginal gyrus. Moreover, patient scores on the Dizziness Handicap Inventory (DHI) score showed a negative correlation with GMV in the posterior insula–operculum regions. Conclusion These findings demonstrated abnormality in the central vestibular cortex and correlations between dizziness severity and GMV in core regions of the vestibular cortex of VM patients, suggesting a pathophysiological role of these core vestibular regions in VM patients.

vestibular symptoms, such as recurrent episodes of vertigo and migrainous symptoms including headache, photophobia, and phonophobia (Dieterich, Obermann, & Celebisoy, 2016;Furman, Marcus, & Balaban, 2013). Epidemiological studies have reported a lifetime VM prevalence of 1% and a 1-year prevalence of 0.89% in the general population (Hochman, Preter, Neuhauser, & Lempert, 2001;Neuhauser et al., 2006). Women suffer VM twice to three times as frequently as men (Lempert & Neuhauser, 2009). Recurrent vertigo and migraine headaches impose a significant burden on the individual as well as society, resulting in productivity loss and even disability.
Gray matter volume (GMV) changes have been detected in several other vestibular disorders, such as vestibular neuritis (VN) and persistent postural perceptual dizziness (PPPD). Some studies have described structural alterations in patients with VN, which may be related to central vestibular compensation. Eulenberg et al.
In contrast, some studies on voxel-based morphometry (VBM) have shown that the GMV is increased regionally in the frontal, occipital, and angular regions in VM patients compared to controls (Messina et al., 2017;Wang et al., 2019). Furthermore, correlation analyses have suggested that VM may induce cumulative effects on the structure of the brain. These analyses show a reduction in GMV during pain, and the vestibular processing areas may be associated with longer disease duration and increased headache severity (Obermann et al., 2014). Based on the above-mentioned structural findings, recurrent VM attacks ultimately result in morphological alterations in the brain areas that are involved in pain and vestibular processing. However, these studies have some limitations. The patient clinical parameters examined in such studies were mainly focused on the pain intensity of migraine attacks and clinical measures of headache-related disability via the Headache Impact Test-6 (HIT-6), Migraine Disability Assessment (MIDAS), and visual analogue scale (VAS). However, these studies did not consider any questionnaires measuring dizziness-related symptoms (Obermann et al., 2014). According to previous studies, many VM patients suffer from balance problems, including vertigo attacks (Akdal, Baykan, et al., 2015;Akdal, Ozge, & Ergor, 2013Cho et al., 2015;Dieterich et al., 2016).
Previous structural and functional imaging has suggested that the parieto-insular vestibular cortex (PIVC), located in the posterior insula, retroinsular region, and parietal operculum, is the core of the human vestibular cortex and plays a key role in vestibular processing (Dieterich & Brandt, 2018;Lopez, Blanke, & Mast, 2012;Ventre-Dominey, 2014;zu Eulenburg, Caspers, Roski, & Eickhoff, 2012). Therefore, in this study, we hypothesized that GMV might change in the PIVC regions of patients with VM compared to normal controls (NCs) and that these changes might be correlated with patients' clinical parameters. We used the Computational Anatomy Toolbox (CAT12) to assess whether GMV changed in patients with VM. Next, a correlation analysis was performed to identify previously unreported structural brain changes associated with migraine and dizziness attacks. To avoid limitations from previous studies, our study not only considered migraine-related clinical questionnaires but also incorporated a dizziness-related clinical questionnaire, the Dizziness Handicap Inventory (DHI), to assess the severity of this symptom in patients with VM. MRI scans were performed on day 3-7 after a VM attack, and all patients were required to be free of migraines and vertigo on the experimental day. All patients underwent a routine neurologic and neuro-otological examination. No peripheral vestibular dysfunction was found in videonystagmography (VNG) recordings. Demographic data were collected from the patients in a face-to-face interview with a standardized questionnaire and questions. Patients rated the pain intensity of migraine attacks using the VAS (0 = no pain; 10 = worst possible pain) and additionally completed the MIDAS, HIT-6 and DHI (Balci, Senyuva, & Akdal, 2018;Sauro et al., 2010). Four patients with VM were on migraine-preventive medications (e.g., beta-blockers).

| Subjects
Additionally, six patients used nonsteroidal analgesics for attack treatment. Most of the investigated patients (n = 10) did not take any medication regularly.
Twenty age-, sex-, education-, and handedness-matched (righthanded) NCs from the community with no history of migraine; chronic pain; previous VN; Meniere's disease; secondary somatoform vertigo; substance abuse; neurologic, mental, or systemic disorders; ischemic or hemorrhagic stroke; or severe head trauma were included. None of these subjects showed structural abnormalities or visible T2-weighted hyperintensities in deep white matter on MRI examination. This study was approved by the Ethics Committee of the Shaanxi Provincial People's Hospital. All participants voluntarily provided written informed consent forms before entering the study.

| Imaging data acquisition
The structural data were acquired on a 3.0 T Philips Ingenia scanner using a 16-element phased-array with only a head coil. A highresolution 3D magnetization-prepared rapid-acquisition gradient echo (MPRAGE) T1-weighted sequence covering the whole brain
CAT12 is one of the most important neuroimaging analysis techniques used to assess structural differences in regional gray volume (Besteher et al., 2017). Moreover, CAT12 can avoid operational bias in the selection of brain regions and in automated measurement of the whole brain. This toolbox includes bias-field and noise removal; skull stripping; segmentation into gray matter, white matter, and cerebrospinal fluid; and, finally, normalization to MNI space using diffeomorphic anatomical registration using exponentiated Lie algebra (DARTEL) to a 1.5 mm isotropic adult template provided by the CAT12 toolbox. The resulting images were checked for homogeneity. As all the images had high correlation values (>0.85), none of the images had to be discarded. Finally, the gray matter images were smoothened using a Gaussian kernel with a full width at half maximum (FWHM) of 8 mm.

| Statistical analysis
The group differences in demographic variables were examined by using independent t tests and analysis of covariance (ANCOVA) in SPSS 22.0. GMV was compared between VM patients and NCs using two-sample t tests in SPM12 with patients' age, sex, and total intracranial volume (TIV) as covariates. The results were assessed at a threshold of p < .05 (false discovery rate [FDR] corrected) with a minimum cluster size set of 100 voxels. Thereafter, the statistically significant brain regions of the VM group were extracted as regions of interest (ROIs), and their correlation with patients' clinical parameters (including the severity of the headache attacks, disease duration [month], number of days per month with headaches [n], MIDAS, HIT-6, and DHI) was analyzed using Pearson's partial correlation analysis in SPSS 22.0. Age and gender were controlling for as covariates. The significance threshold was set at p < .05.

| Clinical data
The clinical and demographic characteristics of the VM and NC groups are presented in Table 1. There were no significant differences in age, sex, or years of education between VM patients and NCs (p > .05; Table 1).

| Correlation of clinical parameters with GMV
The results of the correlation analysis showed that the DHI scores were negatively correlated with the volume of the left posterior insula-operculum regions (p = .042, r = −0.459; Figure 2).

| D ISCUSS I ON
In the present study, a significant decrease in GMV was observed in patients with VM compared to NCs. The main brain regions that showed a significant decrease in GMV were the PFC, the posterior insula-operculum regions, the inferior parietal gyrus, and the supramarginal gyrus. Correlation analysis revealed that the reduction of GMV in the posterior insula-operculum regions was negatively correlated with the severity of dizziness. There is evidence that abnormality of the central vestibular cortex is involved in the In this study, the approaches were confined to a priori defined brain regions. Therefore, it is impossible to locate the widespread patterns of abnormalities across the brain in VM patients. In contrast, Messina and colleagues showed that VM patients had a selective increase in GMV in the frontal lobe, thalamus, temporal lobe, and occipital lobe when compared to NCs. These inconsistent results might be explained by demographic inconsistencies, such as white matter hyperintense lesions, in some VM patients who might affect the structural changes in the brain. Our study enrolled VM patients without white matter hyperintense lesions. Meanwhile, approaches in our analysis used voxel-wise whole-brain methods to avoid any unintentional bias by a priori hypotheses.
The striking finding of our study was the reduction of GMV in the . The other major finding of this study was the reduced GMV in the operculum, which was adjacent to the posterior insula. The posterior insular-opercular regions are believed to be contribute to pain transmission as the receiving areas of the spinothalamic system, which remains as the crucial part of the pain network (Frot, 2003;Garcia-Larrea, 2012;Isnard et al., 2011;Mazzola et al., 2012). Therefore, our findings of GMV loss in these regions might reflect the important role of these transmission circuitry impairments in the pathophysiology of VM. Furthermore, GMV in the posterior insula-operculum regions was negatively correlated with the severity of vertigo in VM patients. From animal and human studies, the vestibular cortex includes the posterior insula, the superficial part of the temporoparietal junction and the superior temporal region, the sensorimotor cortex, the hippocampus, and many other structures of the parietal, frontal, and occipital lobes (Kahane, Hoffmann, Minotti, & Berthoz, 2003).  (Frank, Wirth, et al., 2016). The studies suggest that PIVC and PIC, although adjacent to each other, play different roles in the integration of visual and vestibular signals (Frank & Greenlee, 2014;Frank, Sun, et al., 2016;Frank, Wirth, et al., 2016). However, they are key regions of the cortical vestibular network. These areas are regarded as the core regions for receiving vestibular information and signal processing (Dieterich & Brandt, 2018;Ventre-Dominey, 2014;zu Eulenburg et al., 2012). Lesion studies have shown that damaged insula area affects the perception of verticality or causes vertigo (Brandt, Botzel, Yousry, Dieterich, & Schulze, 1995;Halgren, Walter, Cherlow, & Crandall, 1978). Therefore, structural impairments in the posterior insula-operculum regions might result in central vestibular syndromes that manifest along with vertigo and dizziness.
The PFC is one of the most prominent areas associated with brain abnormalities in patients with migraine (Rocca et al., 2014;Schmitz et al., 2008;Schwedt & Dodick, 2009). Previous studies have suggested that the PFC plays a key role in connecting limbic and subcortical areas and is deemed to be associated with pain perception, modulation of pain, and cognitive and emotional variables (Apkarian, Baliki, & Geha, 2009;Apkarian, Bushnell, Treede, & Zubieta, 2005;Wiech, Ploner, & Tracey, 2008). Obermann and colleagues (Obermann et al., 2014) found decreased GMV in the PFC of VM patients compared to NCs using VBM methods. Our results also revealed that multiple areas within the PFC, including the middle frontal gyrus (MFG) and orbital medial frontal gyrus (OMFG), were altered in VM patients. One study suggested that structural PFC deficits, especially abnormalities in the MFG regions of the brain, might be closely related to the monitoring and temporal organization impairments in VM patients (Fassbender et al., 2004). According to a recent meta-analysis of VBM results in patients with migraine, a significant GMV reduction in the middle gyrus was observed (Hu, Guo, Chen, & He, 2015). These findings are consistent with the above-mentioned studies, suggesting that these areas were involved in cortical processing of vestibular and nociceptive information.
Additionally, several other regions with decreased GMV were observed, including the inferior parietal lobule and supramarginal gyrus.
Alterations in the GMV of the inferior parietal lobule have been reported in many previous cerebral structural studies on migraine (Yu et al., 2016). Those studies showed that the cortical thickness of the inferior parietal lobe in the migraine patient group was significantly decreased compared with that in the healthy control group. Previous fMRI studies during vestibular stimulation in healthy subjects indicated a complex network of brain areas that are involved in the equilibrium and spatial navigation (Bottini et al., 2001). Among these areas, the inferior parietal lobule has been involved in maintaining attention when working toward the current task goals and responding to the salient new information or alerting stimuli in the environment (Singh-Curry & Husain, 2009). A previous study showed activation of the inferior parietal lobule in VM patients (Teggi et al., 2016). The inferior parietal lobule mainly functions in the pain response and the sensing of temperature and pressure. This region is also part of the multisensory vestibular cortical network .
Our findings in VM patients suggested that it might be associated

| Limitations
Several limitations of the current study may require some consideration. First, the number of VM patients is relatively small, and the necessity for additional study in examining large samples may help elucidate the apparent volumetric changes in VM. Second, preventive therapies that might influence the brain morphological results could not be excluded. Third, migraine subtypes such as migraine without aura (MWoA) and migraine with aura (MWA) were not identified. Finally, our study merely utilized the methodology for examining GMV in VM patients. Future studies might combine structural and functional analyses to help us better understand the pathophysiology of VM.

| CON CLUS IONS
In summary, the current study showed alterations in the vestibular cortex of patients with VM. Some of these findings, particularly in the posterior insula-operculum, involve vestibular cortical core regions that play a pathophysiological role in patients with VM.

ACK N OWLED G M ENT
This work was supported by the Shaanxi Provincial Science and Technology Development Funds (2017-032).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interests.

AUTH O R CO NTR I B UTI O N S
Xia Zhe drafted the manuscript, study concept or design, and statistical analysis. Li Chen undertook clinical parameters assessments.
Fuxia Bai, Ze Zou, Xin Zhang, and Weibo Chen provided technical support. Jie Gao, Min Tang, Dongsheng Zhang, Xuejiao Yan, and Xiaoyan Lei acquisition of data. Xiaoling Zhang study supervision or coordination. All authors read and approved the final manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on requests from the corresponding author.