The main findings of this cross-sectional study, which was conducted in a secondary headache clinic, can be summarized as follows: 1) Of the 245 patients, 65 (26.5%), 74 (30.2%), and 106 (43.3%) patients were assigned to the VM group, MwVS group, or MwoVS group, respectively. 2) Variables with significantly different values among the three groups were female gender, migraine with aura, migraine with visual aura, migraine intensity, HIT-6 score, prevalence of depression, migraine with tinnitus, prevalence of sleep disorders, migraine with CA, and migraine with IWPH. 3) Post-hoc pairwise comparison revealed that the VM group was significantly associated with aura, visual aura, headache disability, depression, tinnitus, sleep disorders, CA, and IWPH compared with the MwoVS group. The comparison between the VM and MwVS groups showed no significant differences in variables, except for the HIT-6 score. 4) The frequency of all CA subtypes showed significant differences among the three groups. The VM group was significantly more strongly associated with all CA subtypes compared with the MwoVS group. The MwVS group was significantly more strongly associated with thermal CA compared with the MwoVS group. Between the VM and MwVS groups, there was no significant difference in the frequency of CA and IWPH.
Vestibular migraine prevalence
Vestibular symptoms were identified in 56.7% of the migraine patients who participated in this study, which was consistent with the proportion observed in previous studies, while dizziness was observed in 51.7–61% of migraine patients [29–31]. VM was observed in 26.5% of the migraine patients in this study, which is more than twice as high as the 9–12% rate reported before the new criteria were developed [2, 29] and the 10.3% reported after the new criteria were established . Calhoun et al. reported that there was a strong correlation between migraines and subjective complaints of vertigo, and that nearly half of the subjects suffered from vertigo or dizziness when the headache intensity was 7 or higher . Based on this result, the high prevalence of VM in this study may be due to selection bias according to the study population used and clinical context at a secondary headache clinic.
Differences between the VM and MwVS groups
There were no significant differences in migraine-related variables between the VM and MwVS groups except for the HIT-6 score. It is reasonable that the VM group, which consists of patients with moderate to severe vertigo, showed high disability (i.e., high HIT-6 score) compared with the MwVS group, which consists of patients with mild vertigo, as HIT-6 is a measure of headache disability. Abouzari et al. compared the clinical features of migraine patients with vestibular symptoms between the following two cohorts: the VM cohort corresponding to the VM group in this study and the migraine headache cohort equivalent to the MwVS group in this study . They concluded that the differences between cohorts represent selection bias rather than meaningful features that are unique to the cohorts . Our study supported their conclusion that VM and MwVS may be on a spectrum of the same disease processes. Therefore, to elucidate the VM pathophysiology, it may be appropriate to compare the group that combines the VM group and the MwVS group with the MwoVS group.
Female gender and vestibular migraine
Our results showed that the VM group had a female preponderance, which was consistent with a recent report . Moreover, a female preponderance was observed in both the MwVS and MwoVS groups. Female gender was significantly different among the three groups in this study; however, multiple comparisons did not reveal a significant difference. This result could have been caused by an overall female ratio of 81.2%, which is a highly biased variable. Increasing the number of cases is expected to make a significant difference. Studies on experimentally induced pain have reported that women exhibited greater pain sensitivity, enhanced pain facilitation, and reduced pain inhibition compared with men . The female gender has been reported to be associated with CA [10, 26, 36, 37]. A higher susceptibility of CA may be influenced by the relationship between female gender and VM.
Aura and vestibular migraine
In previous studies, migraine with aura was reported in 13–36% of VM [2, 38]. In this study, the rate of migraine with aura (56.9%) was considerably high. However, this difference was considered to be due to selection bias caused by our clinic providing neurosurgery, thus attracting a higher number of patients with this migraine with aura. In Japan, many migraine patients with aura tend to undergo neurosurgery, since visual aura is considered a stroke warning.
The correlation between vertigo and migraine with or without aura is epidemiologically controversial. Previous studies found an association between migraine with aura and vertigo [31, 39], whereas other studies have reported that patients with migraine without aura have vertigo more frequently [3, 38, 40–42]. A prospective, cross-sectional study documented a significantly high prevalence of vertigo/dizziness (24.5% vs. 12.1%) in migraine with aura compared with migraine without aura (p < 0.01) . Consistent with this view, we found a strong association between migraine with aura and the VM group. In the present study, post-hoc pairwise comparisons revealed that aura was significantly associated with the VM group compared with the MwoVS group (p = 0.007). In addition, visual aura was significantly associated with the VM group compared with the MwoVS group (p = 0.049). Meanwhile, there was no significant difference in migraine with aura (p = 1.000) and visual aura (p = 1.000) between the MwVS and MwoVS groups. These results provide a reasonable basis for assuming that aura plays an important role throughout the VM stage. Vestibular symptoms in VM patients can be explained by cortical spreading depression (CSD) . Demarquay et al. suggested that the vestibular aura (vertigo/dizziness) be classified as a typical migraine aura, which is the clinical consequence of the CSD-induced transient dysfunction of the parieto-insular vestibular cortex . The significant association between the VM group and visual aura suggests that vestibular aura might develop when CSD from the occipital lobe propagates to the adjacent vestibular cortex, or by the opposite propagation. Interestingly, aura but not photophobia was significantly associated with the VM group, can be explained by the involvement of the cuneus and lingual gyrus in photophobia  and the possible origin of a typical visual aura in V3A . In addition, aura is reportedly associated with CA [11, 36, 37]. Our results show that acute CA was significantly associated with the VM group compared with the MwoVS group. Therefore, aura may play an important role in VM.
Headache intensity and vestibular migraine
There have been few studies on the association between VM and headache intensity. Kutay et al. reported no significant difference in headache between a VM group and a migraine without vertigo group . Conversely, Calhoun et al. reported a strong correlation between migraine pain and the subjective complaint of vertigo, and headache intensity of 7 or greater was concomitant with vertigo or dizziness in almost half of the subjects . Furthermore, headache intensity was found to be associated with CA [11, 37]. In this study, headache intensity was highest in the VM group and was significantly different among the three groups. However, no significant difference was found between each pair in the multiple comparison, which may be due to an insufficient number of cases showing statistical significance. Vestibular nuclei receive a projection from the trigeminal nucleus , and trigeminal nerve stimulation causes nystagmus in patients with migraine, which indicates increased vestibular severity excitability compared with the healthy controls . Recently, Zhang et al. demonstrated the sensitization of vestibular nucleus neurons for impairing vestibular function in an induced chronic migraine rat model . Stronger trigeminal nerve inputs may have contributed to stronger sensitization of vestibular nuclei. Therefore, the significant difference in headache intensity among the three groups in this study as well as the highest intensity being observed in the VM group were reasonable. The pathophysiology of VM vestibular symptoms may involve a modulation of vestibular signals through vestibular pathways, including vestibular nuclei, which are sensitized by severer headaches.
Headache disability and vestibular migraine
The HIT-6 score is a reliable and validated tool for assessing headache-related disability. In this study, a post-hoc comparison revealed that the HIT-6 score was significantly high in the VM group compared with the MwVS group and MwoVS group. Intuitively, the HIT-6 score may be not a variable that is related to VM pathophysiology, but it may have been enhanced by the impact of vertigo-associated migraine. The ICHD-III beta criteria for VM with moderate to severe vestibular symptoms lasting between 5 minutes and 72 hours will identify patients who have a more debilitating type of migraine. In addition, headache disability has been reported to be associated with CA [10, 11, 37].
Depression and vestibular migraine
The prevalence of depression was significantly different among the three groups. A post-hoc comparison revealed that the prevalence of depression was significantly higher in the VM group than in the MwoVS group. Migraine, vestibular disorder, and psychological disorders such as anxiety and depression are all closely related [47, 51, 52]. Furman et al. conceptualized this overlap of migraine, anxiety, and balance disorder as migraine-anxiety related dizziness (MARD). The pathophysiology of MARD probably relates primarily to monoaminergic pathways, which are important for migraine, anxiety, the central vestibular system, and their interconnections . VM may be a subtype of MARD in which vestibular disorders predominate the three components. Depression has also been correlated with CA in migraine disease [11, 18, 36, 54].
Tinnitus and vestibular migraine
Recent studies have reported tinnitus in 35.3–52.6% of VM patients [55–57]. In this study, tinnitus was found in 43.1% of VM patients, which did not conflict with previous reports. Tinnitus was significantly associated with VM and MwVS groups compared with the MwoVS group. There was no significant difference in the frequency of tinnitus between the VM and MwVS groups. Tinnitus is related to aberrant neural activity at certain levels of the auditory system , which makes it difficult to identify the sites that are responsible for tinnitus. Several potential mechanisms, including the cortical pathway, subcortical pathway, and peripheral pathway, have been proposed for the pathophysiological interaction between tinnitus and migraine.
Possible cortical pathway responsible for tinnitus
Phonophobia may be associated with cortical hypersensitivity in the auditory cortex, which is involved in the processing of sound . Our results show that there was no significant difference in the frequency of phonophobia between the VM and MwoVS groups. Therefore, it seems unlikely that tinnitus is due to hypersensitivity originating from the auditory cortex. In patients with VM, tinnitus may occur when CSD extends secondarily to the auditory cortex.
Possible subcortical pathway responsible for tinnitus
The medial geniculate body, which is a potential target for tinnitus treatment , may be associated with tinnitus in VM patients. Despite having a primary role in auditory signaling from the cochlear nucleus, the medial division of the medial geniculate body receives projections from the spinal cord  and vestibular nuclei [62, 63]. Therefore, in VM patients, the medial geniculate body may receive spinal cord input of amplified somatic pain represented by CA and/or IWPH. Moreover, the medial geniculate body may receive input from vestibular nuclei sensitized by headache input from the trigeminal nerve . When the auditory cortex receives modified auditory inputs via the medial geniculate body sensitized by somatic pain and vestibular inputs, VM patients may be aware of tinnitus. Previous studies have suggested that in approximately 6–12% of normal-hearing persons with tinnitus and spontaneous otoacoustic emissions (SOAEs), the SOAEs are partly responsible for tinnitus . In VM patients, SOAEs that are modulated through the sensitized subcortical auditory pathway, including the medial geniculate body, may be involved in tinnitus.
Possible peripheral pathway responsible for tinnitus
The trigeminovascular system innervates the inner ear . Trigeminal terminals on the cochlear artery may release the calcitonin gene-related peptide  and substance P  in a manner that is similar to what occurs in the dural artery during a migraine . This process promotes vasodilation and plasma extravasation via inflammatory mediators  and may be related to endolymphatic hydrops in the inner ear, thus resulting in vertigo, and tinnitus. These pathways may play a role in the vertigo, hypoacusis, and tinnitus experienced by some migraineurs during their migraine attacks, and which can mimic Meniere’s disease . Since ~ 50% of patients with Meniere’s disease also meet the criteria for migraine [71, 72], tinnitus could be related to a specific subtype of Meniere’s disease . Volcy et al. found a temporal association between the onset and improvement of CA and tinnitus, which suggests that both tinnitus and CA may share a similar pathophysiological mechanism. They assumed that abnormal auditory cortex functioning due to central and/or peripheral sensitization may contribute to the development of tinnitus in patients with migraine, and that tinnitus may be another allodynic symptom in some patients . According to this sensitization hypothesis, tinnitus may be a migraine symptom.
Sleep disorders and vestibular migraine
Consistent with previous studies [75, 76], sleep disorders were significantly high in the VM group compared with the MwoVS group. Sleep disorders can trigger VM headaches, as well as migraine headaches [2, 77]. Allodynia is strongly related to sleep quality, in a bi-directional way; sleep disturbances may favor central sensitization, and, in turn, allodynia may impair sleep .
Allodynia and vestibular migraine
Our results revealed that CA was significantly associated with the VM group. To date, studies on the association between VM and allodynia have been sparse. Based on the ICHD-II criteria, Akdal et al. revealed that allodynia occurred significantly more frequently in migraine patients with vestibular symptoms (n = 663) than in migraine patients without vestibular symptoms (n = 208) (p = 0.001) . Abouzari et al. documented no significant differences in allodynia frequency between definite VM (n = 104) diagnosed by the ICHD-III beta criteria and non-VM migraine (n = 100) patients (p = 0.350) . To the best of our knowledge, this is the first report that is based on the ICHD-III beta criteria to reveal that all CA subtypes (cephalic, extracephalic, thermal, and mechanical CA) were significantly more strongly associated with VM patients compared with MwoVS patients. In brief, VM was strongly associated with “widespread multimodal allodynia.” The development of cephalic CA indicates sensitization of the second-order spinal trigeminal nucleus, whereas extracephalic CA may reflect sensitization of third-order trigeminovascular neurons in the posterior thalamus . Using functional magnetic resonance imaging assessment of blood oxygenation level-dependent signals, Burstein et al. revealed that the spread of multimodal allodynia beyond the locus of the headache is mediated by sensitized thalamic neurons that process nociceptive information from the cranial meninges together with sensory information from the head and whole body . Radiological investigations in humans have demonstrated multiple projections from vestibular nuclei to the thalamus. . We assume that, in VM, vestibular inputs modulated through the thalamus, which are sensitized by repeated headaches that form the vestibulo-thalamo-cortical pathway, are conducted through the cortex and cause vestibular symptoms.
Determining the presence or absence of widespread multimodal allodynia may be supportive of VM diagnosis. The emphasized relationship between CA and VM can be interpreted in different ways. Central sensitization induced by a recurrent headache can simultaneously induce a distortion of cutaneous sensitivity (CA) and spatial perception (vertigo). Alternatively, recurrent stimuli of vertigo in subjects with hypersensitivity to vestibular symptoms may, in cooperation with repetitive headaches, induce central sensitization. Previous studies reported that the average time between the onset of headache and the onset of vestibular symptoms is 8 to 19 years [39, 80]. Given that migraines begin earlier than migrainous vertigo in most patients [2, 42], it is reasonable to assume that thalamic sensitization due to headaches is part of the pathophysiology underlying vestibular symptoms of migraine. In a similar speculation as that of Volcy et al., who assumed that tinnitus would be an allodynic symptom , we assume that abnormal vestibular cortex functioning due to thalamic sensitization may contribute to the vertigo development in patients with migraine and that vertigo may be another allodynic symptom in some patients.
IWPH and vestibular migraine
To date, this is the first report revealing a significantly stronger association between VM patients and IWPH compared with MwoVS patients. The pathophysiology of interictal widespread thermal and pressure hyperalgesia may involve the brainstem descending pain modulatory system [18, 81]. Dysregulation of the descending pain modulatory system, which may underlie the pathophysiology of IWPH , might play an important role in amplifying headache input to the thalamus, thus accelerating thalamic sensitization. Sensitized thalamus may develop into a widespread multimodal CA. IWPH with acute CA, which are clinical manifestations of central sensitization, as evidenced by increases in frequency that were correlated with each other .
Candidate brain region of vertigo in migraine
Based on the above discussion, we hypothesized that vertigo in VM consists of three distinct origins: 1) vertigo originating cortically, which may be a possible focal symptom due to CSD; 2) vertigo originating subcortically, which may be related to the modulation of vestibular input by the sensitized thalamus; 3) vertigo originating peripherally, which may be related to Meniere’s disease-like inner ear disorders. Among them, clinical manifestation of thalamic sensitization, which is a widespread multimodal CA, was specifically associated with VM, which indicates that thalamic sensitization may play a key role in VM pathogenesis. The presence of IWPH may be involved in susceptibility to thalamic sensitization, as it may indicate a brain state that facilitates noxious trigeminal input (headache) to reach the thalamus.
Strengths and limitations
The strengths of this study include a large cohort size, well-defined migraine status, detailed information on migraine-associated symptoms, and detailed descriptions about allodynia. An objective finding that was associated with the impaired descending pain modulatory system, IWPH, was investigated in a practical and reproducible manner using a manual tender point survey without quantitative sensory testing. However, this study had limitations. First, patients were recruited from a secondary headache clinic seeking further medical therapies for headache rather than vertigo. Thus, refractory patients who visited a tertiary headache center or patients with mild headaches that improved with conventional therapies were likely excluded from the cohort. In addition, migraine patients who visited an otolaryngology department with a chief complaint of moderate to severe vertigo may have been excluded from this study. The headache intensity and disability of our cohort may have been biased toward a homogenized moderate or more severe state, while those of vertigo may have been biased toward a homogenized moderate or less severe state, compared to those who were treated in the general population. This biased population may have limited the generalizability of the results and conclusions. Second, because our data were collected through a questionnaire for the headache population, recollection bias regarding the nature and intensity of vestibular symptoms is a potential limitation. Third, only one rater of IWPH was involved, which may have caused the limitation in of a lack of inter-rater reliability. Fourth, due to the cross-sectional design, we can only describe the association of symptoms. Therefore, we cannot draw firm conclusions about the potential causal relationship between migraine-associated symptoms and vestibular symptoms that meet the diagnostic criteria for VM.
Although factors associated with VM in this study can be generalized for patients who visited a secondary headache clinic rather than the general population, the demographic and clinical features of migraine, including prevalence of vestibular symptoms, were consistent with those of previous studies that were conducted in several countries. Further studies are needed to corroborate our findings.
Directions for future research
Future research should focus on evaluating whether the results could be replicated by other raters in evaluating cohorts consisting of other nationalities and ethnicities. In collaboration with findings in the fields of otolaryngology and neurology, further research should be conducted to elucidate the relationship between detailed allodynia and detailed vestibular symptoms. Findings that can be obtained with clinical examinations, including widespread multimodal allodynia/hyperalgesia for patients with vertigo, may provide quality clues, and insights into VM pathophysiology through the lens of central sensitization.