Circadian variations in occurrence and the clinical presentation of Vestibular Migraine: A retrospective study


 Background: As a new clinical disease entity, vestibular migraine (VM) is considered to be the commonest cause of spontaneous episodic vertigo. This study is trying to explore the circadian variations in occurrence and the clinical presentation of VM.Methods: We conducted a retrospective study and identified female patients who fulfilled the 2012 International Headache Society-Bárány Society Criteria for VM. Population’s clinical data were collected, including onset time and descriptions of sensations experienced by our patients during VM attacks.Results: A total of 189 female VM patients were included in our study. 74.0% of attacks in VM were presented in morning hours before 12 o’clock and the peak of occurrence was at about 7 o'clock. The attack frequency reached the baseline during 12:00-23:59 while there were two abnormal upward fluctuations at about 14:00 and 20:00. The biological circadian cycles may have greater impact of VM than lack of sleep. In addition, clinical presentations of vestibular symptoms including orthostatic vertigo, visually induced vertigo and dizziness showed variations among four 6-hour quadrants per day. Conclusions: Occurrence as well as clinical presentations of vestibular symptoms exhibited circadian variations among VM patients. These data suggested that chronobiological mechanisms may play a role in vestibular migraine pathophysiology.


Introduction
As the most common neurologic cause of episodic vertigo, vestibular migraine (VM) is a variant of migraine resulting in vestibular symptoms in addition to migraine (1,2). Despite a prevalence of between 1% and 2.7% of adult population (3), diagnosis of VM is challenging perhaps due to the broad spectrum of its manifestations, complex temporal patterns and its very recent nosological de nition (4,5). Several biological phenomena and diseases have speci c circadian rhythms (6)(7)(8). The relatively high occurrence of migraine or benign paroxysmal positional vertigo attacks have been widely documented upon awakening in recently studies (9,10). We also noticed that a lot of VM patients experienced sudden vertigo with or without severe headache during the morning hours in our clinic practice. Therefore, we hypothesized that the occurrence and the presentation of VM may vary during different circadian periods. We conducted a review of VM patients of two hospitals to assess circadian variations in occurrence and the clinical presentations of vestibular symptoms.

Study Population
We conducted a retrospective review of patients at the neurology outpatient clinics of the First A liated Hospital of Soochow University and Suzhou TCM Hospital A liated to Nanjing University of Chinese Medicine. A total of 197 female patients who ful lled the 2012 International Headache Society-Bárány Society Criteria (included in the third version of the International Classi cation of Headache Disorder (ICHD-3)) for VM or probable VM were followed up by three months (11)(12)(13). The diagnoses were made by two senior neurologists and one senior otolaryngologist.
The exclusion criteria were as follows: (1) headache or vestibular symptoms attributed to secondary causes; (2) other causes of vestibular attack such as benign paroxysmal positional vertigo, Meniere's disease or transient ischemic attack of posterior circulation; (3) users of oral contraceptives or migraine prophylactic; (4) history of alcohol or drugs abuse; (5) history of head trauma or intracranial infection; (6) history of otorhinolaryngology surgery; (7) shift work.

Clinical information collection
We reviewed all study population's data: age, race, job, history of caffeine and alcohol, age of onset, relevant medical history, relevant family history, the onset time of this attack, symptoms experienced during this attack, self-reported triggers for this attack, neuro-otologic examinations, and clinical balance  (11,14). All patients gave informed consent and more detailed methods are available from the corresponding author upon reasonable request.
Continuous variables were analyzed as mean and standard deviation or the median and interquartile range while categorical variables were analyzed as frequency and percentage, properly. Differences among these variables were assessed by the chi-square test or ANOVA. Post hoc analysis was performed with Bonferroni correction. Data analysis were presented using kernel density, radar plots and 100% stacked bar. The level of signi cance for these descriptive comparisons was established at 0.05 for twosided hypothesis testing. Statistical analysis was performed in SPSS 25.0.

Result Demographics and medical history
After 3-month follow-up, we identi ed 189 female VM patients according to the inclusion and exclusion criteria. Demographics and relevant medical history were showed in Table 1. The age was 42.0 (35.0, 52.0) years old; the duration of illness was 4.0 (1.0, 8.0) years. As a well-recognized comorbidity in migraine, 63.5% sufferers experience motion sickness, more than previous studies. 71 patients (37.6%) reported a family history of similar episodic vestibular symptoms (lasting from 5 minutes to a few days).

Clinical presentation
For all 189 attacks, activities at onset were classi ed into moving, rising, sitting and sleeping. Rising (36.5%) was presented most frequently and sitting (10.1%) was presented least. Even most attacks presented during the sleeping time, activities of moving their head around (moving and rising) were more relevant to provoking VM attacks (60.8% VS 29.2%) ( Table 2).
Most our patients experienced at least one vestibular symptom during attacks of VM. Spontaneous vertigo was the predominant vestibular symptom (42.3%) and it's frequencies did not signi cantly differ among four 6-hour quadrants after adjustment. Dizziness was the second most frequent category of vestibular complaints ( 29.1%) and reported more frequently during 12:00-17:59, followed by 18:00-23:59.

Self-reported triggers
Sleep deprivation was frequently reported as a trigger for VM attacks. Figure 4 shows the distribution of attacks (n=189) among four 6-hour quadrants per day, and sleep-related triggers was considered the precipitating factor. The percentage of sleep deprivation was shown in the Figure 3, and there was no signi cant difference among four quadrants.

Discussion
The key ndings in our study are as follows: (1) 74.0% of attacks in VM were presented in morning hours before 12 o'clock; occurrence of VM peaked in the morning hours between 07:00-07:59, and reached valley between 21:00-21:59; (2) clinical presentations of vestibular symptoms showed variations among four 6-hour quadrants per day; (3) attacks may be in uenced by poor sleep, however, the biological circadian cycles may have greater impact of VM than lack of sleep. To our knowledge, this study was the rst time to analyze the circadian variations in the occurrence and the clinical presentation of VM.
In our population, the age of onset was 36.00 (29.50, 44.00) years old, consistent other studies (15). But few of our patients had a personal history of alcohol or caffeine, that may due to the gender or local dietary habit and may affect the composition of in uence factors.
For all attacks, frequencies of activities at onset differed among four 6-hour quadrants and "rising" (36.5%) was presented most frequently. The onset patterns of vestibular symptoms in each quadrant was related to the regular activity patterns of our population. But on the whole, vestibular symptoms were more often triggered by activities of moving their heads, which perhaps was one trigger unique to VM but not migraine headache (4). The attack of VM presented more frequently between 00:00-12:00 and reached its peak about 7 o'clock in the morning. This nding is similar to previous chronobiological studies that reported an increased frequency of morning headache attacks among migraineurs (16). We also found abnormal upward uctuations at 14:00 and 20:00, which may correlate to the taking of food and suggesting a role of the sympathetic nervous system (17).
In addition, most of our VM patients experienced more than one accompanying symptom and vestibular symptom. Orthostatic vertigo was reported more frequent in the morning hours while dizziness was reported more frequent in the afternoon hours. Headaches may or may not accompany vestibular symptoms during VM attacks, consistent with previous study (5). The majority reported nausea with or without vomiting, photophobia or phonophobia. Just a few of our patients described tinnitus and visual aura, which is much lower than those data reported in other studies (4,18).
Through these analyses, we can get a preliminary conclusion about circadian variations of VM attacks. However, what is the possible mechanism in the occurrence and clinical presentation of VM underlying the circadian rhythm? The pathophysiology of VM has not been fully established yet (19,20). In present hypotheses, as a a variant of migraine, VM is the integral overlap among vestibular pathways, migraine circuit triggers and central mechanisms for premonitory symptom generation (21)(22)(23).Hypothalamic activation and circadian variation has been reported during migraine attacks in many recently studies (16,24,25). The circadian rhythm is controlled by a complex system of molecular regulation with a master precursor, located in the suprachiasmatic nucleus of the anterior hypothalamus (26,27). In pathophysiological model of VM by Furman, the hypothalamus was contained within a network of interoceptive circuits for vestibular, visceral sensory, and nociceptive information (28).Therefore, the circadian variations in the occurrence and the clinical presentation of VM may due to the hypothalamic involvement in both nociception and circadian periodicity (28). Similar to the migraine, peak incidence of VM is usually during sleep or upon awakening (9,29). A key question is whether the observed temporal pattern represents a true endogenously mediated circadian pattern, or it is just triggered by alteration of the sleep pattern. However, the proportion of attacks induced by abnormal sleep did not increase signi cantly in VM patients with morning onset. Lack of sleep may indeed be one of the triggers of attacks, but endogenously circadian rhythm of VM maybe dominant.
Our data should be interpreted with some caution due to limitations of the study. These included the relatively small sample size and retrospective bias inherent. Secondly, we enrolled patients with probable VM, who might develop de nite VM over time as some studies have shown (30)(31)(32). Moreover, we only described the clinical features of VM by a cross-sectional study, further study on the relevant mechanism will be great helpful to identify this entity.

Consent for publication
Not applicable.

Availability of data and materials
Original data of the present study are available from the corresponding author upon reasonable request.

Competing interests
All the authors declare no con ict of interest.

Funding
This work was supported by the grants from Youth Science and Technology Project of "Promoting Health through Science and Education" in Suzhou (KJXW2019041).   Figure 1 Kernel density for the circadian variation in VM occurrence. The peak of occurrence was at about 7 o'clock in the morning.